IBM Z NetView

6.4

Tuning Guide

IBM

SC27-2874-06
 Note
Before using this information and the product it supports, read the information in “Notices” on page
141.

This edition applies to version 6, release 4 of IBM Z NetView (product number 5697-NV6 ) and to all subsequent
versions, releases, and modifications until otherwise indicated in new editions.
This edition replaces SC27-2874-05.
© Copyright International Business Machines Corporation 1997, 2022.
US Government Users Restricted Rights – Use, duplication or disclosure restricted by GSA ADP Schedule Contract with
IBM Corp.
Contents

Figures................................................................................................................ vii
About this publication...........................................................................................ix
Intended audience...................................................................................................................................... ix
Publications................................................................................................................................................. ix
IBM Z NetView library............................................................................................................................ ix
Related publications .............................................................................................................................. x
Terminology in this Library.................................................................................................................... xi
Using IBM Z NetView online help.......................................................................................................... xi
Accessing publications online............................................................................................................... xi
Ordering publications ...........................................................................................................................xii
Accessibility ............................................................................................................................................... xii
Support information................................................................................................................................... xii
Conventions used in this publication......................................................................................................... xii
Typeface conventions .......................................................................................................................... xii
Operating system-dependent variables and paths.............................................................................xiii
Syntax diagrams...................................................................................................................................xiii

Chapter 1. Improving NetView Performance...........................................................1

Achieving Performance Goals......................................................................................................................1
Using General Techniques........................................................................................................................... 1

Chapter 2. Tuning for Automated Operations.......................................................... 5

Tuning Techniques....................................................................................................................................... 5
Limiting System Messages...........................................................................................................................5
Using EMCS Console Support...................................................................................................................... 6
NetView Subsystem Address Space............................................................................................................7
NetView Automation Table.......................................................................................................................... 7
Using BEGIN/END to Improve Efficiency...............................................................................................7
Using Other Techniques to Improve Efficiency..................................................................................... 8
AUTOCNT Command.............................................................................................................................. 9
Automating Hardware Monitor Records....................................................................................................13
Filtering Hardware Monitor Records......................................................................................................... 13
Automation Tasks (Autotasks)...................................................................................................................13
Using Multiple Autotasks..................................................................................................................... 14
Using Full-Screen Automation............................................................................................................. 14
Command and Message Forwarding......................................................................................................... 15
Separation of the Automation Workload from Other NetView Workloads...............................................15
Multiple NetView Programs .................................................................................................................15
Single NetView Program Using WLM Enclaves.................................................................................... 16

Chapter 3. Tuning for AON....................................................................................19

Tuning Techniques..................................................................................................................................... 19
CNMCMD Resident Option......................................................................................................................... 19
DSICLD Library..................................................................................................................................... 19
Automation Table................................................................................................................................. 19
DDF Tree and Panel.............................................................................................................................. 20
Node Automation................................................................................................................................. 20
Operations..................................................................................................................................................20
TCP/IP Support for AON............................................................................................................................ 21

iii
 Chapter 4. Tuning for Command Procedures......................................................... 23
Tuning Techniques..................................................................................................................................... 23
Command Lists.......................................................................................................................................... 24
Preloading Command Lists.................................................................................................................. 24
Managing Command Lists with AUTODROP........................................................................................ 25
Subroutines.......................................................................................................................................... 25
REXX Command Lists.................................................................................................................................26
How to Compile REXX Procedures.......................................................................................................26
Compiled REXX/370 Command Lists.................................................................................................. 26
REXX Function Packages......................................................................................................................27
Tuning REXX Environments..................................................................................................................30
Command Processors................................................................................................................................31
Command Processors Written in a High-Level Language................................................................... 32
Running High-level Language Programs in a Preinitialized Environment........................................... 32
Global Variables......................................................................................................................................... 34
Enhancing Performance....................................................................................................................... 35
Save/Restore Processing..................................................................................................................... 35

Chapter 5. Tuning for the Hardware Monitor......................................................... 37

Tuning Techniques..................................................................................................................................... 37
Hardware Monitor Filters...........................................................................................................................37
Alerts-Dynamic Panel................................................................................................................................ 39
Using the NPDA.ALCACHE Statement....................................................................................................... 39
Using the NPDA.ALERTLOG Statement..................................................................................................... 41
Using the NPDA.ALERTFWD Statement ................................................................................................... 41
HMSTATS Command.................................................................................................................................. 41
Using the NPDA.DSRBO Statement...........................................................................................................44
Wrap Counts...............................................................................................................................................44
SWRAP Command................................................................................................................................ 44
Initialization Specifications..................................................................................................................45
Error-to-Traffic (E/T) Ratio Thresholds..................................................................................................... 45
SRFILTER and SRATIO Commands......................................................................................................45
RATIO Statement Initialization Specifications.................................................................................... 46
NPDA.RATE Statement Initialization Specifications........................................................................... 46
Event/Automation Service......................................................................................................................... 47

Chapter 6. Tuning for the Session Monitor.............................................................49

Major Tuning Techniques for the Session Monitor.................................................................................... 49
SAW Data....................................................................................................................................................50
SAW Buffer Allocation and Tuning....................................................................................................... 50
Tuning the SAW Buffer Allocation........................................................................................................51
Trace Data.................................................................................................................................................. 52
Global Tracing.......................................................................................................................................52
Selective Tracing.................................................................................................................................. 52
PIU Buffer Allocation and Tuning.........................................................................................................54
KEEPPIU Parameter............................................................................................................................. 55
TRACEGW Parameter........................................................................................................................... 56
Keep Classes.............................................................................................................................................. 56
Network Resource Naming Conventions............................................................................................. 56
SAW Option...........................................................................................................................................57
KEEPPIU Option................................................................................................................................... 58
AVAIL Option........................................................................................................................................ 59
DASD Option......................................................................................................................................... 59
KEEPSESS Option................................................................................................................................. 60
DGROUP Option....................................................................................................................................61
Managing the Session Monitor Database.................................................................................................. 61

iv
 DASD Filtering.......................................................................................................................................61
Managing Database Size...................................................................................................................... 62
SESSMDIS Command................................................................................................................................ 63
RTM Data Collection...................................................................................................................................66
LUCOUNT Parameter................................................................................................................................. 66

Chapter 7. Tuning for the NetView Management Console.......................................69

Workstation Tuning Techniques................................................................................................................ 69
Storage Estimates................................................................................................................................ 70
Hardware Requirements...................................................................................................................... 70
Client Performance...............................................................................................................................71
Using Background Pictures.................................................................................................................. 71
Status Focal Point to Programmable Workstation Connectivity......................................................... 71
Server-Client Configurations................................................................................................................72
Host Tuning Techniques............................................................................................................................ 72
NETCONV.............................................................................................................................................. 72
DUIGINIT Parameters..........................................................................................................................72

Chapter 8. Tuning for the Resource Object Data Manager...................................... 73

Tuning Techniques..................................................................................................................................... 73
Warm Start and CHKPT Commands.......................................................................................................... 73
RODM Data Sets.........................................................................................................................................74
RODM API Statistics.................................................................................................................................. 74
RODM Cell Pool Statistics.......................................................................................................................... 76
Using the Histogram Data.................................................................................................................... 78
Customization Parameters........................................................................................................................ 79
Programming Recommendations..............................................................................................................79

Chapter 9. Tuning for VSAM..................................................................................81

Tuning Techniques..................................................................................................................................... 81
Local Shared Resources (LSR) and Deferred Write (DFR).........................................................................81
Definitions for LSR and DFR................................................................................................................. 82
Buffer Pool Sizes...................................................................................................................................82
Monitoring VSAM Performance..................................................................................................................84
LISTCAT Command...............................................................................................................................84
VSAMPOOL Command..........................................................................................................................87
VSAM Database Maintenance....................................................................................................................88

Chapter 10. Additional Tuning Considerations...................................................... 91

Tuning Considerations............................................................................................................................... 91
Address Space Dispatch Priority............................................................................................................... 91
Automated Operations Network (AON) Performance Considerations..................................................... 92
Browse....................................................................................................................................................... 92
Canzlog Archiving.......................................................................................................................................92
Canzlog Data Set Characteristics......................................................................................................... 93
Canzlog Data Access............................................................................................................................ 93
Canzlog Archive Storage Requirements.............................................................................................. 93
Command Security ....................................................................................................................................95
Command Statistics...................................................................................................................................96
Data Services Request Blocks (DSRBs).....................................................................................................97
Installation Exits........................................................................................................................................ 98
LOGTSTAT Command................................................................................................................................. 98
LU 6.2 Transport........................................................................................................................................ 99
MAXSESS Keyword.................................................................................................................................. 100
NCCF TRACE Options...............................................................................................................................100
NetView Access from the Web Browser..................................................................................................100
NetView Constants Module (DSICTMOD)............................................................................................... 101

v
 NetView-NetView Communication..........................................................................................................101
NetView Program-to-Program Interface................................................................................................ 101
Network Asset Management Facility.......................................................................................................102
Partitioned Data Set (PDS) Allocation..................................................................................................... 102
Persistent and Nonpersistent LUC Sessions...........................................................................................103
Using Nonpersistent Sessions over Dialed Lines.............................................................................. 103
RESOURCE Command............................................................................................................................. 104
Resource Limits....................................................................................................................................... 105
Keywords for Resource Limits........................................................................................................... 105
Using Resource Limits........................................................................................................................106
Status Monitor STATOPT Filtering........................................................................................................... 107
STEPLIB DD Statements..........................................................................................................................107
TASKMON Command............................................................................................................................... 107
TASKUTIL Command............................................................................................................................... 109
TASKUTIL Command Output............................................................................................................. 110
Calculating Task Utilizations with Two Observations of TASKUTIL.................................................. 112
Suggestions for Using TASKUTIL....................................................................................................... 113
IBM Z NetView Enterprise Management Agent...................................................................................... 114

Chapter 11. Storage Considerations................................................................... 115

Estimating Storage Usage....................................................................................................................... 115
Region Size......................................................................................................................................... 133
Keeping Track of Virtual Storage and Other System Resource Usage................................................... 134
Minimizing Storage Usage....................................................................................................................... 135
Coding RES=N on Command Definition Statements.........................................................................135

Notices..............................................................................................................141
Programming Interfaces..........................................................................................................................142
Trademarks.............................................................................................................................................. 142
Privacy policy considerations.................................................................................................................. 142

Index................................................................................................................ 145

vi
Figures

1. Example: Using a REXX CLIST to Collect Performance Statistics................................................................ 4

2. Message Detail Report................................................................................................................................ 10

3. Message Summary Report.......................................................................................................................... 11

4. CNMSJM11 before Modification to Switch Function Package Search Order.............................................28

5. CNMSJM11 After Modification with No User Function Package Defined.................................................. 29

6. CNMSJM11 After Modification with a User Function Package Defined.....................................................30

7. Sample Output of the HLLENV, TYPE=IBMADPLI,LIST,STATS,RESET Command..................................... 34

8. Sample Output of the HLLENV LIST STATUS TYPE=IBMADC.................................................................... 34

9. Hardware Monitor Database and Filters.....................................................................................................38

10. Example of Output from HMSTATS Command......................................................................................... 42

11. Example of KCLASS and MAPSESS Statements.......................................................................................57

12. Example of KCLASS and MAPSESS Statements Using the KEEPSESS and DGROUP Options................ 61

13. Session Monitor Session and Storage Information Panel........................................................................63

14. RODM Log Record Type 8 for API Statistics (Part 1 of 2).........................................................................75

15. RODM Log Record Type 8 for API Statistics (Part 2 of 2).........................................................................76

16. RODM Log Record Type 8 for Segment and Window Statistics............................................................... 77

17. Histogram Data......................................................................................................................................... 78

18. Sample BLDVRP Macros Defining VSAM Buffer Pools for CNMSJM01....................................................83

19. Sample LISTCAT Command Output Using 3390 DASD............................................................................85

20. Sample Output from the VSAMPOOL Command Using 3390 DASD........................................................87

21. Sample Output from the DSRBS Command............................................................................................. 97

22. Sample DISPPI Command Output......................................................................................................... 102

23. Sample Output from the RESOURCE Command.................................................................................... 104

vii
 24. Sample TASKMON Output (Part 1 of 2).................................................................................................. 108

25. Sample TASKMON Output (Part 2 of 2).................................................................................................. 109

26. TASKUTIL Command Output.................................................................................................................. 110

27. Output from TASKUTIL Command, 2 Hours Later................................................................................. 112

viii
About this publication
The IBM Z® NetView® product provides advanced capabilities that you can use to maintain the highest
degree of availability of your complex, multi-platform, multi-vendor networks and systems from a single
point of control. This publication, the IBM Z NetView Tuning Guide, provides tuning information and
techniques for the NetView product. For the purpose of this publication, tuning is activity that helps
the NetView program and its network environment achieve a certain performance goal in areas such as
response time, resource utilization, and throughput. Although this publication does not provide a precise
method of tuning the NetView product, it highlights the areas that have the most effect on NetView
performance and explains how you can improve performance in those areas.

Intended audience
This publication is for system programmers whose jobs involve improving the performance of the NetView
program. Readers should have a thorough understanding of the NetView program. They must also have
some problem diagnosis experience and an idea of what can cause a performance goal to be missed.

Publications
This section lists publications in the IBM Z NetView library and related documents. It also describes how
to access NetView publications online and how to order NetView publications.

IBM Z NetView library

The following documents are available in the IBM Z NetView library:
• Administration Reference, SC27-2869, describes the NetView program definition statements required
for system administration.
• Application Programmer's Guide, SC27-2870, describes the NetView program-to-program interface
(PPI) and how to use the NetView application programming interfaces (APIs).
• Automation Guide, SC27-2846, describes how to use automated operations to improve system and
network efficiency and operator productivity.
• Command Reference Volume 1 (A-N), SC27-2847, and Command Reference Volume 2 (O-Z), SC27-2848,
describe the NetView commands, which can be used for network and system operation and in
command lists and command procedures.
• Installation: Configuring Additional Components, GC27-2851, describes how to configure NetView
functions beyond the base functions.
• Installation: Configuring the NetView Enterprise Management Agent, GC27-2853, describes how to
install and configure the IBM Z NetView Enterprise Management Agent.
• Installation: Getting Started, GI11-9443, describes how to install and configure the base NetView
program.
• Installation: Migration Guide, GC27-2854, describes the new functions that are provided by the current
release of the NetView product and the migration of the base functions from a previous release.
• IP Management, SC27-2855, describes how to use the NetView product to manage IP networks.
• Messages and Codes Volume 1 (AAU-DSI), GC27-2856, and Messages and Codes Volume 2 (DUI-IHS),
GC27-2857, describe the messages for the NetView product, the NetView abend codes, the sense
codes that are included in NetView messages, and generic alert code points.
• Programming: Pipes, SC27-2859, describes how to use the NetView pipelines to customize a NetView
installation.

© Copyright IBM Corp. 1997, 2022 ix

 • Programming: REXX and the NetView Command List Language, SC27-2861, describes how to write
command lists for the NetView product using the Restructured Extended Executor language (REXX) or
the NetView command list language.
• Security Reference, SC27-2863, describes how to implement authorization checking for the NetView
environment.
• Troubleshooting Guide, GC27-2865, provides information about documenting, diagnosing, and solving
problems that occur in the NetView product.
• Tuning Guide, SC27-2874, provides tuning information to help achieve certain performance goals for the
NetView product and the network environment.
• User's Guide: Automated Operations Network, SC27-2866, describes how to use the NetView
Automated Operations Network (AON) component, which provides event-driven network automation,
to improve system and network efficiency. It also describes how to tailor and extend the automated
operations capabilities of the AON component.
• User's Guide: NetView, SC27-2867, describes how to use the NetView product to manage complex,
multivendor networks and systems from a single point.
• User's Guide: NetView Enterprise Management Agent, SC27-2876, describes how to use the NetView
Enterprise Management Agent.
• Licensed Program Specifications, GC31-8848, provides the license information for the NetView product.
• Program Directory for IBM Z NetView US English, GI11-9444, contains information about the material
and procedures that are associated with installing the NetView product.
• Program Directory for IBM Z NetView Japanese, GI11-9445, contains information about the material and
procedures that are associated with installing the NetView product.
• Program Directory for IBM Z NetView Enterprise Management Agent, GI11-9446, contains information
about the material and procedures that are associated with installing the IBM Z NetView Enterprise
Management Agent.
The following books are archived:
• Customization Guide, SC27-2849, describes how to customize the NetView product and points to
sources of related information.
• Data Model Reference, SC27-2850, provides information about the Graphic Monitor Facility host
subsystem (GMFHS) and MultiSystem Manager data models.
• Installation: Configuring Graphical Components, GC27-2852, describes how to install and configure the
NetView graphics components.
• Programming: Assembler, SC27-2858, describes how to write exit routines, command processors, and
subtasks for the NetView product using assembler language.
• Programming: PL/I and C, SC27-2860, describes how to write command processors and installation exit
routines for the NetView product using PL/I or C.
• Resource Object Data Manager and GMFHS Programmer's Guide, SC27-2862, describes the NetView
Resource Object Data Manager (RODM), including how to define your non-SNA network to RODM and
use RODM for network automation and for application programming.
• User's Guide: NetView Management Console, SC27-2868, provides information about the NetView
management console interface of the NetView product.

Related publications
You can find additional product information on the IBM Z NetView web site at https://www.ibm.com/
products/z-netview.
For information about the NetView Bridge function, see Tivoli NetView for OS/390® Bridge Implementation,
SC31-8238-03 (available only in the V1R4 library).

x IBM Z NetView: Tuning Guide

Terminology in this Library
The following terms are used in this library:
CNMCMD
For the CNMCMD member and the members that are included in it using the %INCLUDE statement
CNMSTYLE
For the CNMSTYLE member and the members that are included in it using the %INCLUDE statement
DSIOPF
For the DSIOPF member and the members that are included in it using the %INCLUDE statement
IBM® Tivoli® Netcool®/OMNIbus
For either of these products:
• IBM Tivoli Netcool/OMNIbus
• IBM Tivoli OMNIbus and Network Manager
MVS™
For z/OS® operating systems
MVS element
For the base control program (BCP) element of the z/OS operating system
NetView
For the following products:
• IBM Z NetView version 6 release 4
• IBM Z NetView version 6 release 3
• IBM Tivoli NetView for z/OS version 6 release 2 modification 1
• NetView releases that are no longer supported
PARMLIB
For SYS1.PARMLIB and other data sets in the concatenation sequence
VTAM®
For Communications Server - SNA Services
Unless otherwise indicated, topics to programs indicate the latest version and release of the programs. If
only a version is indicated, the topic is to all releases within that version.
When a topic is made about using a personal computer or workstation, any programmable workstation
can be used.

Using IBM Z NetView online help

The following types of IBM Z NetView mainframe online help are available, depending on your installation
and configuration:
• General help and component information
• Command help
• Message help
• Sense code information
• Recommended actions

Accessing publications online

IBM posts publications for this and all other products, as they become available and whenever they
are updated, to the IBM Documentation at https://www.ibm.com/docs/en. You can find IBM Z NetView
documentation on IBM Z NetView Documentation.
Note: If you print PDF documents on other than letter-sized paper, set the option in the Print window that
enables Adobe Reader to print letter-sized pages on your local paper.

About this publication xi

Ordering publications
You can order many IBM publications online at http://www.ibm.com/e-business/linkweb/publications/
servlet/pbi.wss
You can also order by telephone by calling one of these numbers:
• In the United States: 800-426-4968
• In Canada: 800-879-2755
In other countries, contact your software account representative to order IBM publications. To locate the
telephone number of your local representative, perform the following steps:
1. Go to http://www.ibm.com/e-business/linkweb/publications/servlet/pbi.wss.
2. Select your country from the list and click the grey arrow button beside the list.
3. Click About this site to see an information page that includes the telephone number of your local
representative.

Accessibility
Accessibility features help users with a physical disability, such as restricted mobility or limited vision, to
use software products successfully. Standard shortcut and accelerator keys are used by the product and
are documented by the operating system. Refer to the documentation provided by your operating system
for more information.
For additional information, see the Accessibility appendix in the User's Guide: NetView.

Support information
If you have a problem with your IBM software, you want to resolve it quickly. IBM provides the following
ways for you to obtain the support you need:
Online
Please follow the instructions located in the support guide entry: https://www.ibm.com/support/
home/pages/support-guide/?product=4429363.
Troubleshooting information
For more information about resolving problems with the IBM Z NetView product, see the IBM Z
NetView Troubleshooting Guide. You can also discuss technical issues about the IBM Z NetView
product through the NetView user group located at https://groups.io/g/NetView. This user group is for
IBM Z NetView customers only, and registration is required. This forum is also monitored by interested
parties within IBM who answer questions and provide guidance about the NetView product. When a
problem with the code is found, you are asked to open an official case to obtain resolution.

Conventions used in this publication

This section describes the conventions that are used in this publication.

Typeface conventions
This publication uses the following typeface conventions:
Bold
• Lowercase commands and mixed case commands that are otherwise difficult to distinguish from
surrounding text
• Interface controls (check boxes, push buttons, radio buttons, spin buttons, fields, folders, icons,
list boxes, items inside list boxes, multicolumn lists, containers, menu choices, menu names, tabs,
property sheets), labels (such as Tip:, and Operating system considerations:)
• Keywords and parameters in text

xii IBM Z NetView: Tuning Guide

 Italic
• Citations (examples: titles of publications, diskettes, and CDs
• Words defined in text (example: a nonswitched line is called a point-to-point line)
• Emphasis of words and letters (words as words example: "Use the word that to introduce a
restrictive clause."; letters as letters example: "The LUN address must start with the letter L.")
• New terms in text (except in a definition list): a view is a frame in a workspace that contains data.
• Variables and values you must provide: ... where myname represents...
Monospace
• Examples and code examples
• File names, programming keywords, and other elements that are difficult to distinguish from
surrounding text
• Message text and prompts addressed to the user
• Text that the user must type
• Values for arguments or command options

Operating system-dependent variables and paths

For workstation components, this publication uses the UNIX convention for specifying environment
variables and for directory notation.
When using the Windows command line, replace $variable with %variable% for environment variables
and replace each forward slash (/) with a backslash (\) in directory paths. The names of environment
variables are not always the same in the Windows and UNIX environments. For example, %TEMP% in
Windows environments is equivalent to $TMPDIR in UNIX environments.
Note: If you are using the bash shell on a Windows system, you can use the UNIX conventions.

Syntax diagrams
The following syntax elements are shown in syntax diagrams. Read syntax diagrams from left-to-right,
top-to-bottom, following the horizontal line (the main path).
• “Symbols” on page xiii
• “Parameters” on page xiv
• “Punctuation and parentheses” on page xiv
• “Abbreviations” on page xiv
For examples of syntax, see “Syntax examples” on page xiv.

Symbols
The following symbols are used in syntax diagrams:

Marks the beginning of the command syntax.

Marks the end of the command syntax.

Indicates that the command syntax is continued on the next line.

Indicates that a statement is continued from the previous line.

|
Marks the beginning and end of a fragment or part of the command syntax.

About this publication xiii

 Parameters
The following types of parameters are used in syntax diagrams:
Required
Required parameters are shown on the main path.
Optional
Optional parameters are shown below the main path.
Default
Default parameters are shown above the main path. In parameter descriptions, default parameters
are underlined.
Syntax diagrams do not rely on highlighting, brackets, or braces. In syntax diagrams, the position of
the elements relative to the main syntax line indicates whether an element is required, optional, or the
default value.
When you issue a command, spaces are required between the parameters unless a different separator,
such as a comma, is specified in the syntax.
Parameters are classified as keywords or variables. Keywords are shown in uppercase letters. Variables,
which represent names or values that you supply, are shown in lowercase letters and are either italicized
or, in NetView help, displayed in a differentiating color.
In the following example, the USER command is a keyword, the user_id parameter is a required variable,
and the password parameter is an optional variable.
USER user_id
password

Punctuation and parentheses

You must include all punctuation that is shown in the syntax diagram, such as colons, semicolons,
commas, minus signs, and both single and double quotation marks.
When an operand can have more than one value, the values are typically enclosed in parentheses and
separated by commas. For a single value, the parentheses typically can be omitted. For more information,
see “Multiple operands or values” on page xv.
If a command requires positional commas to separate keywords and variables, the commas are shown
before the keywords or variables.
When examples of commands are shown, commas are also used to indicate the absence of a positional
operand. For example, the second comma indicates that an optional operand is not being used:

COMMAND_NAME opt_variable_1,,opt_variable_3

You do not need to specify the trailing positional commas. Trailing positional and non-positional commas
either are ignored or cause a command to be rejected. Restrictions for each command state whether
trailing commas cause the command to be rejected.

Abbreviations
Command and keyword abbreviations are listed in synonym tables after each command description.

Syntax examples
The following examples show the different uses of syntax elements:
• “Required syntax elements” on page xv
• “Optional syntax elements” on page xv
• “Default keywords and values” on page xv

xiv IBM Z NetView: Tuning Guide

• “Multiple operands or values” on page xv
• “Syntax that is longer than one line” on page xvi
• “Syntax fragments” on page xvi

Required syntax elements

Required keywords and variables are shown on the main syntax line. You must code required keywords
and variables.
REQUIRED_KEYWORD required_variable

A required choice (two or more items) is shown in a vertical stack on the main path. The items are shown
in alphanumeric order.
REQUIRED_OPERAND_OR_VALUE_1

REQUIRED_OPERAND_OR_VALUE_2

Optional syntax elements

Optional keywords and variables are shown below the main syntax line. You can choose not to code
optional keywords and variables.

OPTIONAL_OPERAND

A required choice (two or more items) is shown in a vertical stack below the main path. The items are
shown in alphanumeric order.

OPTIONAL_OPERAND_OR_VALUE_1

OPTIONAL_OPERAND_OR_VALUE_2

Default keywords and values

Default keywords and values are shown above the main syntax line in one of the following ways:
• A default keyword is shown only above the main syntax line. You can specify this keyword or allow it
to default. The following syntax example shows the default keyword KEYWORD1 above the main syntax
line and the rest of the optional keywords below the main syntax line.
• If an operand has a default value, the operand is shown both above and below the main syntax line. A
value below the main syntax line indicates that if you specify the operand, you must also specify either
the default value or another value shown. If you do not specify the operand, the default value above the
main syntax line is used. The following syntax example shows the default values for operand OPTION=*
above and below the main syntax line.

KEYWORD1 OPTION=*
COMMAND_NAME
KEYWORD1 OPTION= *
KEYWORD2 VALUE1

KEYWORD3 VALUE2

Multiple operands or values

An arrow returning to the left above a group of operands or values indicates that more than one can be
selected or that a single one can be repeated.

About this publication xv

 KEYWORD= (
,

REPEATABLE_OPERAND_OR_VALUE_1

REPEATABLE_OPERAND_OR_VALUE_2

REPEATABLE_OPERAND_OR_VALUE_3

value_n )

Syntax that is longer than one line

If a diagram is longer than one line, each line that is to be continued ends with a single arrowhead and the
following line begins with a single arrowhead.
OPERAND1 OPERAND2 OPERAND3 OPERAND4 OPERAND5 OPERAND6

OPERAND7 OPERAND8

Syntax fragments
Some syntax diagrams contain syntax fragments, which are used for lengthy, complex, or repeated
sections of syntax. Syntax fragments follow the main diagram. Each syntax fragment name is mixed case
and is shown in the main diagram and in the heading of the fragment. The following syntax example
shows a syntax diagram with two fragments that are identified as Fragment1 and Fragment2.

COMMAND_NAME Fragment1

Fragment2

Fragment1
KEYWORD_A= valueA KEYWORD_B KEYWORD_C

Fragment2
KEYWORD_D KEYWORD_E= valueE KEYWORD_F

xvi IBM Z NetView: Tuning Guide

Chapter 1. Improving NetView Performance

This chapter addresses many tuning-related questions asked by users of the NetView program. This
chapter also provides basic tuning considerations to improve general NetView performance. Review these
topics and considerations for relevancy to your environment.

Achieving Performance Goals

Tuning is any activity that helps the NetView program and its network environment achieve a certain
performance goal in areas such as response time, resource utilization, and throughput. This goal or
expectation can be as formal as a service level agreement or as informal as a person's perception.
Regardless of the formality of the performance goal, tuning is any activity performed to meet that goal.
The task of tuning is difficult to define. People have many different perceptions and approaches to tuning.
Unlike the installation task, no set sequence of steps leads to a tuned system. This book describes
the areas that affect the performance of the NetView program and shows methods for controlling and
optimizing that performance. It explains how the following factors affect NetView program tuning:
• Filter use
• Automation activity
• Network size
• VSAM data set design, placement, and management
• Command list use, design, and data set placement
• View size
• Workstation CPU and RAM
• Link speed
An important part of tuning is setting the performance goal. However, that goal varies from installation to
installation. Some installations have more message traffic or network activity than others. Many factors
uniquely define the performance goal for each installation; therefore, these factors are not discussed in
this book.
This book explains how to determine certain tuning values and suggests values you should use. Each
environment needs to be tuned according to the following factors:
• Automation levels
• Logon and logoff rates
• Monitoring activities
• Network component failure rates
• Network transaction rates
• Problem determination activities
• System transaction rates
• Resource status change activities
All recommendations are based on experiences and measurements done in an IBM controlled
environment. The use of these recommendations is a customer responsibility and depends on the
customers' ability to evaluate and integrate them into their operational environments.

Using General Techniques

If you are just getting started in tuning the NetView program, use the following general tuning techniques
to improve overall NetView performance. You can handle the most frequently encountered performance
and tuning considerations by using these techniques.

© Copyright IBM Corp. 1997, 2022 1

 1. Use the RATE statement to stop rapidly recurring hardware monitor alerts from being recorded to the
hardware monitor database. See “NPDA.RATE Statement Initialization Specifications” on page 46 for
more information.
2. Decide which data is critical to keep for the session monitor, and set the following session monitor
options and parameters accordingly:
• Sense code values in DSICTMOD for DASD filtering
• KCLASS DASD option
• KCLASS AVAIL option
• KCLASS SAW option
• KCLASS KEEPPIU option
• KCLASS KEEPSESS option

See Chapter 6, “Tuning for the Session Monitor,” on page 49 for more information.
3. Use the DBAUTO command to reorganize or reset the VSAM databases regularly. See “ VSAM Database
Maintenance” on page 88 for more information.
4. Use the following utilities to monitor performance:
AUTOCNT
Generates automation table usage reports. See “AUTOCNT Command” on page 9.
COLLCTL LISTINFO
Displays the average and maximum time it takes for the NetView program data collectors to collect
data during a given data collection interval. The display includes statistics for the following data
collectors:
• DVDEF
• DVTAD
• DVCONN
• DVROUT
• INTERFACES
• TELNET
• APPL
DISPPI
Displays information about NetView program-to-program interface buffer queues, including buffer
queue lengths, total buffers sent, and buffer storage usage. See “NetView Program-to-Program
Interface” on page 101.
DSRBS
Displays statistics on data services request block (DSRB) use for NetView and user-written data
services tasks (DSTs). See “Using the NPDA.DSRBO Statement” on page 44.
HLLENV
Displays information about usage of the regular and critical preinitialized environment pools
by PL/I command processors. See “Running High-level Language Programs in a Preinitialized
Environment” on page 32.
HMSTATS
Displays hardware monitor event and alert workload counters, and statistics about the alert cache
and alerts dynamic (ALD) screen update processing. See “HMSTATS Command” on page 41.
LISTCAT
Displays VSAM database definition and performance data for NetView DSTs that have open VSAM
databases. See “LISTCAT Command” on page 84.
LOGTSTAT
Logs record type 38 subtype 2 to SMF at logoff/logon/termination.

2 IBM Z NetView: Tuning Guide

 MAPCL
Provides information about preloaded CLISTs. See “Preloading Command Lists” on page 24.
NACTL LISTINFO
Displays the average and maximum time it takes for the IBM Z NetView Enterprise Management
Agent data collectors to collect data during a given data collection interval. The display includes
statistics for the following agent subtowers:
• HEALTH
• CONNACT
• CONINACT
• SESSACT
QRYGLOBL
Displays information about NetView global variables, including the expected number of common
or task global variables and the actual number of variables found. See “Global Variables” on page
34.
RESOURCE
Provides CPU and storage statistics for the NetView address space. See “RESOURCE Command” on
page 104.
SESSMDIS
Displays session monitor session counts, storage use, and workload traffic information. See
“SESSMDIS Command” on page 63.
STATAPI
Generates RODM API statistics to analyze the content and activity of RODM. See “RODM API
Statistics” on page 74.
STATCELL
Collects information concerning the distribution of storage cells in windows and segments. See
“RODM Cell Pool Statistics” on page 76.
TASKMON
Shows CPU, storage, I/O, and message queue rate statistics for all tasks in the NetView program.
TASKURPT
A NetView sample that demonstrates how reports can be generated from the task resources data
in the SMF log using REXX.
TASKUTIL
Displays task performance information, including CPU utilization, queue lengths, storage use, and
active command lists. See “TASKUTIL Command” on page 109.
VSAMPOOL
Displays VSAM local shared resource (LSR) pool use statistics. See “VSAMPOOL Command” on
page 87.

You can add these utilities to a command list that can be run periodically to collect performance statistics.
See the example in Figure 1 on page 4.

Chapter 1. Improving NetView Performance 3

 /* REXX clist to collect performance statistics. Can be invoked by an*/
/* EVERY timer to run under an autotask - for example, */
/* EXCMD AUTO2,EVERY 00:15,V621PERF */
/* Substitute your RODMNAME for X below! If you're not using RODM, */
/* delete the lines referring to X below. */
/* */
/* */
'MVS F X,STATAPI,CLEAR'
'AUTOCNT REPORT=BOTH,STATS=SUMMARY'
'COLLCTL LISTINFO'
'DISPPI'
'DSRBS AAUTSKLP'
'DSRBS BNJDSERV'
'DSRBS DSIGDS'
'HMSTATS'
'LISTCAT AAUTSKLP'
'LISTCAT BNJDSERV'
'LISTCAT DSILOG'
'MAPCL'
'MVS F X,STATCELL'
'NACTL LISTINFO'
'RESOURCE'
'SESSMDIS'
'TASKMON * *'
/* 'TASKURPT (NOWINDOW' */
'TASKUTIL'
'VSAMPOOL'
say 'End of V64PERF CLIST'
exit

Figure 1. Example: Using a REXX CLIST to Collect Performance Statistics

Recording performance statistics to the NetView log in this manner can provide useful historical
information for trend analysis and can be invaluable in performance problem analysis.

4 IBM Z NetView: Tuning Guide

Chapter 2. Tuning for Automated Operations

The NetView program provides functions that you can use to automate operations.
Many messages are sent to the NetView program for automation after being processed by the MVS
message processing facility (MPF).
For a detailed description of NetView automation, refer to theIBM Z NetView Automation Guide.

Tuning Techniques
Following are the major tuning techniques for automated operations, arranged in order of expected effect
on performance, with the most important tuning considerations listed first. These are described in detail
in this chapter.
1. Limit the number of system messages processed by the NetView program. On MVS systems, include
a .NO_ENTRY statement specifying AUTO(NO) in the MPFLSTxx member of SYS1.PARMLIB. See
“Limiting System Messages” on page 5.
2. Segment the automation table using BEGIN/END sections to minimize the average number of
automation table statements evaluated for each message or MSU. See “Using BEGIN/END to Improve
Efficiency” on page 7.
3. Use the AUTOCNT command to generate usage reports for the NetView automation table. See
“AUTOCNT Command” on page 9.
4. Eliminate simple command procedures when the condition processing can be performed directly in the
automation table. See “Using Other Techniques to Improve Efficiency” on page 8.
5. Block the hardware monitor ESREC, AREC, and OPER filters where possible for alerts that are to be
automated. See “Filtering Hardware Monitor Records” on page 13.
6. Minimize the use of the CONTINUE automation table action if it causes the entire table to be scanned.
See “Using Other Techniques to Improve Efficiency” on page 8.
7. Use the MSUSEG automation table condition if you plan to automate alerts. This automates the MSU
directly, rather than generating the BNJ146I and BNJ030I messages using the hardware monitor
OPER filter. See “Automating Hardware Monitor Records” on page 13.
8. Use multiple autotasks for MVS environments to distribute the automation workload across multiple
processors. See “Automation Tasks (Autotasks)” on page 13.
9. Use the RMTCMD command, where possible, to forward commands and messages between a
distributed system and a focal point. The alternative method uses OST-NNT sessions. See “Command
and Message Forwarding” on page 15.

Limiting System Messages

To use NetView automation support with the NetView program, the first and most important step is to
limit the number of system messages the NetView program processes. The NetView automation table is
searched for most messages that pass through the NetView program. Therefore, it is important to limit the
system message traffic sent to the NetView program to only those messages that should be considered
for automation or those that are needed for display purposes.
When you automate the operation of distributed hosts with a focal point host, handle as much of the
traffic as possible at the distributed host to minimize the overhead of forwarding messages to the focal
point host for automation.
You can use the NetView Revision Table function to make finely-tuned decisions about the following:
• Which messages are passed to NetView for automation

© Copyright IBM Corp. 1997, 2022 5

 • Which messages are eligible for further z/OS processing, including the overhead of routing to other parts
of the sysplex.
Of the actions available in the Revision Table, the following three are of particular interest for limiting
performance cost of system messages:
AUTOMATE
Set this to OFF for messages that are not needed by your automation tables in the NetView program.
DELETE
Set this to ON for messages that are not needed anywhere.
NETVONLY
Set this to ON for messages that should only be automated and not seen, logged, or routed, or until
automation specifies otherwise.
Other Revision actions can also have performance benefits. For example, you might want to limit
the routing codes set in some messages. See the IBM Z NetView Automation Guide and also sample
CNMSMRT1.
To guard against a message flood that causes a backlog in automation, you can use z/OS Message
Flood Automation (MFA). Messages that are acted upon by MFA get special consideration in NetView
Message Revision. Such messages that are not marked for automation will trigger action under
UPON(FLOOD=NOAUTO). While these messages do not generally cause trouble for NetView automation,
you may want to examine such messages in the more particular way supported by the NetView MRT and
take different actions. Also, the presence of an UPON(FLOOD=NOAUTO) will prevent other UPON actions
from triggering and prevent an unintended action during a message flood.
For messages that are acted upon by MFA, but still deemed automatable when submitted to Message
Revision, the UPON(FLOOD=AUTO) will trigger for them and allow further decisions to be made at that
point.
If you are not using the Revision Table, use the MPFLSTxx member of SYS1.PARMLIB to identify whether a
message is eligible for automation processing. Eligible messages are passed to the NetView program over
the subsystem interface through the Canzlog service. (see “NetView Subsystem Address Space” on page
7).
The NetView subsystem will make a decision about passing messages to the NetView program after
MPFLSTxx processing. Consider to specify the following items for MPFLSTxx:
• Specify AUTO(YES) or AUTO(NO) to identify whether a message is eligible for automation processing.
• Specify the default values for groups of messages listed in the MPFLSTxx member with the .DEFAULT
statement. The .DEFAULT statement uses AUTO(NO) unless you specify AUTO(YES).
• Specify the default processing you want for messages that are not identified in the MPFLSTxx member
with the .NO_ENTRY statement.
Note: Ensure that your MPFLSTxx member has a .NO_ENTRY statement with AUTO(NO) specified. This
helps limit the number of messages sent to the NetView program to only those needed for automation
purposes.
The .NO_ENTRY statement uses AUTO(YES) unless you specify AUTO(NO). If you do not have
a .NO_ENTRY statement in MPFLSTxx, the system uses a default of AUTO(YES).
Note: When using a Revision Table, its actions follow and supercede actions by the MPFLSTxx process.
For more information about MPF, refer to the appropriate MVS publication.

Using EMCS Console Support

The NetView program uses extended multiple console support (EMCS) consoles. However, the NetView
subsystem transfers messages for automation to the CNMCSSIR task independently of EMCS consoles.
You can assign attributes to the consoles and change the attributes of your EMCS consoles to ensure
that messages with certain route codes are delivered to the consoles you specify. See the IBM Z NetView
Security Reference for a description of the EMCS console attributes.

6 IBM Z NetView: Tuning Guide

NetView Subsystem Address Space
The NetView subsystem address space job must be running to enable you to:
• Start the program-to-program interface
• Enable the NetView Message Revision function.
Table 8 on page 125 describes how to calculate the minimum, maximum, and recommended values for
use in the NetView subsystem address space job.

NetView Automation Table

Automation of system and network messages and MSUs is accomplished with automation statements
(IF-THEN, END, ALWAYS, SYN, and %INCLUDE) in the NetView automation table, which is activated by
the AUTOTBL command. Messages and MSUs are passed through the active NetView automation table to
sequentially search one or more of the following automation table statements that indicate actions to be
automatically performed for that message or MSU.
• IF-THEN contains conditions to be tested and actions to be taken if the conditions are satisfied. Actions
that can be taken include a direct response to a system message, a command or command procedure to
be run, or presentation modification. Multiple actions can be specified on a single IF-THEN statement.
• END indicates the end of a BEGIN/END section that can be used to logically segment automation
statements.
• ALWAYS indicates that certain actions are always to be done if the sequential search through the active
automation table reaches the ALWAYS statement.
• SYN defines table synonyms that represent values that can subsequently be substituted within other
automation table statements.
• %INCLUDE physically segments the table into multiple members or files.

If an automation table is activated as a result of a NetView AUTOTBL command, a compact, preprocessed

version of the automation table is loaded into NetView storage. This allows a quicker search of the active
NetView automation table than if the actual automation members are scanned when a message or MSU
is processed. SYN and %INCLUDE statements do not affect the processing time of messages and MSUs.
These statements are processed when the AUTOTBL command is processed, not when each message or
MSU is processed.
You can decrease the processing time for messages and MSUs by following specific automation table
design guidelines. The guidelines in the rest of this section decrease the average search time for the
NetView program to locate matching statements and to perform actions.
As you make changes to improve the efficiency of the automation table, you should monitor the
performance of the table using the AUTOCNT command. See “AUTOCNT Command” on page 9 for
more information.

Using BEGIN/END to Improve Efficiency

• Because the NetView automation table is searched sequentially (beginning to end), logically segmenting
the automation table using BEGIN/END sections provides a large performance advantage. Using
BEGIN/END reduces the number of automation statements that must be scanned. You should design
your automation table so that each message prefix (such as DSI), each MSU major vector (such as major
vector X'0000'), and specific subvectors have BEGIN/END sections.
If you dedicate BEGIN/END sections to the message and MSU classes, the number of statements that
must be scanned is smaller than if all of the IF-THEN and ALWAYS statements are listed. If each
message class has its own unique BEGIN/END section, placing ALWAYS as the last statement in a
section prevents the automation table from being scanned further if no automation table match is
found.

Chapter 2. Tuning for Automated Operations 7

 • Order BEGIN/END sections by frequency of use. If a large number of VTAM messages are received, a
BEGIN/END section to handle IST messages can be at or near the beginning of the automation table to
decrease the number of statements that must be evaluated before the correct BEGIN/END section is
found.
• Order the statements within BEGIN/END sections by frequency of use. Minimizing the number of
statements that must be searched to find the correct matching statement reduces processing.

Using Other Techniques to Improve Efficiency

The following guidelines are other techniques you can use to decrease the average search time for the
NetView program to locate matching statements and to perform actions.
• Place statements for entire classes of messages or MSUs that are frequently received, but not
processed near the beginning of the table or BEGIN/END sections. IF-THEN statements and ALWAYS
statements can specify that no action be performed, eliminating unnecessary processing time. For
example, if you do not automate commands issued at a NetView terminal, you can add the statement in
the following example at the beginning of your table:

IF HDRMTYPE = '*' THEN ;

• Some comparison items take longer to evaluate than others. Comparison items with the potential to
be relatively slow include MSUSEG comparison items that specify complex locations, the DSICGLOB
automation table function (ATF) program that is supplied with the NetView product, and any lengthy ATF
programs you have written. Isolate these items by placing them in BEGIN/END sections started with an
IF-THEN statement so that the NetView program evaluates the items only when the comparison in the
IF-THEN statement is true. You can also isolate items by placing them after a logical-AND operator (&).
In this case, the NetView program evaluates the items only if the conditions before the & are met.
• Avoid unnecessary NetView automation processing by using the operating system message facility
(MPF for MVS) to suppress unneeded messages. Only those system messages that require NetView
automation or an operator's action should be forwarded to the NetView program. See “Limiting System
Messages” on page 5 for more information.
• Eliminate simple command procedures whose function can be processed in the automation table
directly. Consider the following examples:
– Instead of calling a command list from the automation table to check the value of a NetView global
variable, you can bypass the processing needed to run the command list by checking the global
variable value directly in the automation table using the DSITGLOB or DSICGLOB ATFs.
– The THRESHOLD comparison item is useful for specifying particular actions to take place when a
condition has happened at least a specified number of times within a specified time period. Using
THRESHOLD in the automation table is more efficient than calling a command procedure to make a
similar determination. The THRESHOLD counters are reset when a new automation table is loaded.
– If you call a command procedure to evaluate a complex condition that the existing automation table
comparison items cannot address, consider writing your own ATF to perform this processing.

• Minimize use of the CONTINUE automation table action that scans the entire table. The CONTINUE
action is useful for things such as setting defaults for the table or a section of the table and then
continuing the search for an additional match, as specified by the statement in the following example:

ALWAYS SYSLOG(Y) NETLOG(N) DISPLAY(Y) CONTINUE(Y) ;

If specified on every statement, CONTINUE(Y) increases processing time. The entire automation table
must be searched to determine all automation actions that should take place. If you use CONTINUE(Y)
to determine multiple IF-THEN statements that all result in actions to be performed, design your table
so that scanning does not need to continue through the entire table by doing one of the following:
– Specify CONTINUE(N) on the last statement where you expect to find a match for a particular
message or MSU.

8 IBM Z NetView: Tuning Guide

 – Specify ALWAYS as the last statement of the BEGIN/END section to prevent continued statements
from being scanned outside of an enclosing BEGIN/END section.

• If automation slows because many unsolicited messages are queued on a single task waiting for
automation table processing, you can use the ASSIGN command to split the messages among several
tasks. If you use the ASSIGN command, you can still use the automation table for final routing of the
message; the ASSIGN command just gets the message to the automation table faster.
For more information about the NetView automation table, message routing, and ASSIGN processing,
refer to IBM Z NetView Automation Guide.

AUTOCNT Command
The AUTOCNT command produces a report describing the usage of automation table statements in the
active NetView automation table. You can also use the AUTOCNT command to reset the automation table
usage counters. You can display the report as multiline messages or place the information in a file.
The AUTOCNT command can request information and statistics on message-type automation statements,
on MSU-type automation statements, or both. You can request summary information or detail. The
detailed information tells you how many messages and MSUs were compared to each automation table
statement, and how many matched.
Message CNM493I is written to the network log each time a match is found in the NetView automation
table that results in the scheduling of a command or command list for execution with the EXEC action.
This message identifies the statement that ran the command or command list. Use it to determine how
frequently commands are run from the automation table and by which automation table statements. If
you use the AUTOCNT command to analyze automation table activity, you might want to prevent the
CNM493I message from being written to the network log. You can use the CNM493I parameter of the
DEFAULTS and OVERRIDE commands or the CNM493I action in the automation table to control whether
the message is written.
If you use the AUTOCNT command to analyze automation table activity, you might want to prevent the
CNM493I message from being written to the network log. You can use the CNM493I parameter of the
DEFAULTS and OVERRIDE commands or the CNM493I action in the automation table to control whether
the message is written.
For syntax and other information about the AUTOCNT command, and for information about the CNM493I
action of the automation table, refer to the IBM Z NetView Automation Guide. For information on the
CNM493I option of the DEFAULTS and OVERRIDE commands, and on the CNM493I message, refer to the
NetView online help.

Detail Reports
Figure 2 on page 10 illustrates the output of a detail report for MSG-type automation table statements.
Detail reports for MSU-type automation table statements provide the same types of data.

Chapter 2. Tuning for Automated Operations 9

 - DWO800I AUTOMATION TABLE MSG DETAIL REPORT BY OPER1

DWO803I ------------( AUTOSEG1 MESSAGE DETAILS 03/30/13 14:32:42 )------------

DWO805I |-- PERCENTAGES --|
DWO806I STMT SEQ MEMBER COMPARE MATCH E C A MATCH/ COMP/ MATCH/
DWO807I NUMBER NUMBER NAME COUNT COUNT C I I COMP TOTAL TOTAL
DWO808I -----------------------------------------------------------------------
DWO809I 00001 00000800 AUTOSEG1 2304 798 34.6 100.0 34.6
DWO809I 00002 00001000 AUTOSEG1 798 177 22.2 34.6 7.7
DWO809I 00003 00001400 AUTOSEG1 621 9 1 1.4 27.0 0.4
DWO809I 00004 00001600 AUTOSEG1 612 0 1 0.0 26.6 0.0
DWO809I 00005 00002000 AUTOSEG1 612 612 X 100.0 26.6 26.6
DWO809I 00007 00002700 AUTOSEG1 1506 160 10.6 65.4 6.9
DWO809I 00008 00002900 AUTOSEG1 160 52 32.5 6.9 2.3
DWO809I 00009 00003400 AUTOSEG1 108 1 0.9 4.7 0.0
DWO809I 00010 00003700 AUTOSEG1 107 107 X 100.0 4.6 4.6
DWO808I ----------------------------------------------------------------------

Figure 2. Message Detail Report

In addition to the statement number, sequence number, and member name, the detail report contains the
following information for each automation table statement.
• Conditional comparisons (COMPARE COUNT)
The counter that increments when the associated conditional statement is selected for evaluation.
• Evaluation matches (MATCH COUNT)
The counter that increments when the associated conditional statement is evaluated as true, resulting
in execution of all automation actions specified on the statement.
• Executed commands (E C)
This column reports the number of commands that are run for this automation statement when there is
an evaluation match. If the number of EXEC actions with CMD keywords is greater than 99, an asterisk
(*) appears in the column.
• Continue indicator (C I)
A report column marked X indicates that the conditional statement contained a CONTINUE action,
causing the NetView program to continue to scan the automation table. CONTINUE(Y) actions cause
additional conditional processing for later statements in the table, and can enable a conditional match
on additional statements.
• Always statement indicator (A I)
A report column marked X indicates that the statement was an ALWAYS. For ALWAYS statements, the
MATCH/COMP field is always 100%.
• Match to compare percentage (MATCH/COMP)
A statistic calculated by dividing the ratio of MATCH COUNT by the COMPARE COUNT of the conditional
statement, multiplied by 100. If the number of matches and the number of comparisons are both zero,
the ratio is shown as -.- to indicate division by zero.
• Compare percentage (COMP/TOTAL)
A statistic calculated by dividing the ratio of COMPARE COUNT of the conditional statement by the total
number of messages or MSUs, multiplied by 100. If the number of comparisons against this statement
and the total number of messages or MSUs processed by automation are both zero, the ratio is shown
as -.- to indicate division by zero.
• Match percentage (MATCH/TOTAL)
A statistic calculated by dividing the ratio of MATCH COUNT of the conditional statement by the total
number of messages or MSUs, multiplied by 100. If the number of matches for this statement and the
total number of messages or MSUs processed by automation are both zero, the ratio is shown as -.- to
indicate division by zero.
Any numeric column value that exceeds 99999999 is overwritten with eight asterisks (*).

10 IBM Z NetView: Tuning Guide

Suggestions for Using Detail Reports
• The report output represents a snapshot of the automation table data, and that automation might
be in progress for some messages or MSUs while the data is being collected. There might be minor
discrepancies between totals in a summary report and values in a detail report produced at the same
time.
• The percentage fields can be interpreted intuitively.
– COMP/TOTAL is the percentage of messages or MSUs processed through the table that compare
against this statement. The first statement always has a COMP/TOTAL percentage of 100%, because
all messages or MSUs compare against the first statement.
– MATCH/TOTAL is the percentage of messages or MSUs processed through the table that match on
this statement.
– COMP/MATCH is the percentage of messages or MSUs compared against this statement that match.
• Smaller COMP/TOTAL percentages indicate better performance. You can reduce the COMP/TOTAL
percentages by using BEGIN/END sections and by ordering statements within sections by match
frequency or MATCH/TOTAL percentage. In other words, statements with higher MATCH/TOTAL
percentages should be located early in their BEGIN/END sections. Likewise, BEGIN-END sections with
higher MATCH/TOTAL percentages should be located early in the automation table.
• COMP/MATCH percentages of 100% are expected on ALWAYS statements. The ALWAYS indicator field
(A I) is included in the report so that you can spot these without referring to a LISTING file. If you
have other statements with a COMP/MATCH percentage of 100%, you should determine whether the
statement is coded correctly.
• The executed commands field (E C) multiplied by the MATCH COUNT field equals the number
of commands run as a result of matches on the automation table statement. This product (E C
times MATCH COUNT), summed for all automation table statements, represents the total number of
commands run from the automation table. This total is shown as the TOTAL COMMANDS EXECUTED
field in the summary report.
• Examine statements with high MATCH COUNT values that execute commands.
– Use the LOADCL command to preload command lists that are executed frequently.
– For command processors that are executed frequently, ensure that RES=N is not coded on the
CMDDEF statement in initialization member CNMCMD.
For more information on making command lists and command processors resident, see “Preloading
Command Lists” on page 24 and “Command Processors” on page 31.

Summary Reports
Figure 3 on page 11 illustrates the output of a summary report for MSG-type automation table
statements. Summary reports for MSU-type automation table statements provide the same types of data,
except for TOTAL ROUTES EXECUTED, because the ROUTE keyword of the EXEC action is not applicable to
MSUs.

- DWO801I AUTOMATION TABLE MSG SUMMARY REPORT BY OPER1

DWO810I ------------( AUTOSEG1 MESSAGE SUMMARY 03/30/13 14:32:42 )-------

DWO812I STATISTICS STARTED = 03/30/13 13:32
DWO813I TOTAL MSGS PROCESSED = 2304
DWO814I MSGS MATCHED = 958
DWO815I MSGS RESULTING IN COMMANDS = 9
DWO816I TOTAL COMMANDS EXECUTED = 9
DWO817I TOTAL ROUTES EXECUTED = 1
DWO818I AVERAGE COMPARES/MSG = 2.58
DWO819I TOTAL MSGS/MINUTE = 38
DWO820I MINUTES ELAPSED = 60
DWO808I ------------------------------------------------------------------

Figure 3. Message Summary Report

The summary report contains the following information.

Chapter 2. Tuning for Automated Operations 11

 • Date and time of usage report generation.
The date is in the format mm/dd/yy. The time is in the format hh:mm:ss, where hh is based on a 24
hour clock. The date and time are reported in the label message for the SUMMARY statistics (messages
DWO810I and DWO811I).
• Date and time of start of usage count monitoring.
The date is in the format mm/dd/yy. The time is in the format hh:mm:ss, where hh is based on a 24 hour
clock. The date and time are reported in message DWO812I.
• Total number of messages or MSUs processed.
A count of all the messages or MSUs that have passed through the automation table.
• Total number of messages or MSUs matched.
The number of messages or MSUs that were acted upon by at least one automation table statement. An
ALWAYS statement causes a message or MSU to be considered a match.
• Number of messages or MSUs resulting in command execution.
A count of the number of messages or MSUs that resulted in one or more commands being executed
from automation table statements.
• Total commands executed for messages or MSUs.
The total number of commands executed by all automation table statements during the period when
statistics were taken. The EXEC action with the CMD keyword indicates a command is executed from the
automation table.
• Total routes executed for messages.
The total number of routes executed by all automation table statements during the period when
statistics were taken. The EXEC action with the ROUTE keyword (and without the CMD keyword)
indicates a route is executed from the automation table.
• Average number of compares per message or MSU.
The number of compares divided by the number of messages or MSUs that had passed through the
automation table.
• Average number of messages or MSUs processed per minute.
The number of messages or MSUs processed by the NetView automation table divided by the number of
minutes since the last reset or load of the automation table.
• Number of minutes elapsed.
The amount of time, in minutes, since the last AUTOCNT RESET command or since the current active
automation table was activated.

Using Summary Reports

The following suggestions can help you use the information in the summary tables:
• The AVERAGE COMPARES/MSG field or AVERAGE COMPARES/MSU field is a good indicator of the
efficiency of the automation table. Performance improvements to the automation table, such as
increased use of BEGIN/END sections, usually result in a decrease in the AVERAGE COMPARES/MSG
value. Monitor this value before and after making changes to the automation table to ensure that your
changes do not have a negative effect on performance.
• The TOTAL MSGS/MINUTE field or TOTAL MSUS/MINUTE field is the TOTAL MSGS PROCESSED field
divided by the MINUTES ELAPSED field. You can calculate other rates, such as total commands
executed per minute or total routes per minute, by performing the arithmetic on the appropriate fields.
Increased filtering by the operating system message processing facility should reduce the TOTAL MSGS/
MINUTE rate.
• Consider setting an EVERY timer under an autotask to invoke the AUTOCNT command periodically. The
AUTOCNT output can be used as important historical data in spotting trends in automation activity.

12 IBM Z NetView: Tuning Guide

 Because summary reports are shorter than detail reports, you might create them more frequently,
perhaps every hour, to monitor activity rates.
• The automation table statistics are set to zero when an automation table is loaded. Prior to loading a
new table, you might want to issue the AUTOCNT command for the old table to collect its statistics.

Automating Hardware Monitor Records

You can automate problem notifications sent to the hardware monitor by generating messages and
sending the messages to the automation table for processing. Many problem records sent to the hardware
monitor are management service units (MSUs). You can generate messages from the MSUs or automate
MSUs directly using the MSUSEG and HIER conditions in the NetView automation table. Direct automation
is more efficient than automation of records converted to messages.
Refer to IBM Z NetView Automation Guide for more information about automating MSUs.

Filtering Hardware Monitor Records

Filtering unnecessary hardware monitor records prevents processing by the NetView program, VSAM, and
your hardware monitor operators. The following filter types are available:
• The ESREC filter determines whether the record should be sent to the events and statistics database
and whether the record is eligible for the remaining filters.
• The AREC filter determines whether the record should be sent to the alerts database, and thus be made
available to operators viewing the hardware monitor alerts panels. The AREC filter also determines
whether the record is eligible for the remaining two filters.
• The OPER filter determines whether BNJ146I and BNJ030I messages should be created from the alert
for automation.
• The ROUTE filter determines whether the alert should be forwarded to the hardware monitor focal
point.

You can set each of the filter types to BLOCK or PASS. A filter set to BLOCK prevents the filter or any
filters dependent on it from taking effect. For example, an AREC filter set to BLOCK prevents the OPER
and ROUTE filters from receiving the alert.
You can set the hardware monitor recording filters using the hardware monitor SRFILTER command.
You can set filters for MSUs that undergo automation processing using the SRF automation table action.
You can use the automation table to override settings made using SRFILTER. Both SRFILTER and the
automation table can set color and other highlighting attributes.
Filtering unnecessary hardware monitor records from the events and statistics database and the alerts
database can increase the productivity of operators and save substantial processing time. Use the
following guidelines when setting filter options:
• Allow required events to pass the ESREC and AREC filters.
• Set the ROUTE filter to PASS only for alerts that must be passed to a focal point system.
• Set the OPER filter to PASS only for alerts that you want to automate by creating BNJ146I and BNJ030I
messages.
For more information on hardware monitor filters, see “Hardware Monitor Filters” on page 37.

Automation Tasks (Autotasks)

An automation task (autotask) is an operator station task (OST) that is not associated with a terminal.
Autotasks can be used to monitor status and to issue responses to system messages.
Do not use the following under an autotask:
• The &PAUSE statement for the NetView command list language.

Chapter 2. Tuning for Automated Operations 13

 • The PARSE PULL and PARSE EXTERNAL statements for REXX.
• The WAIT FOR OPINPUT statement for high-level languages.
These statements do not provide a timeout facility. If you pause in a command list executed under an
autotask, use a facility that includes a GO command, because the GO command breaks the wait implied
by the paused statement.

Using Multiple Autotasks

Using multiple autotasks has the following advantages:
• Improved reliability
If you have only one autotask and it is logged off or disabled, this can affect the entire automation
process. Having multiple autotasks minimizes the impact of an autotask failure.
• Improved problem determination
For example, if VTAM fails, having an autotask specifically designed for VTAM processing makes it easy
for an operator to determine the autotask recovery stage.
• Improved audit trail
It is easier to determine what work has been done by an autotask (for example, by looking at the
network log) when the work is divided among multiple tasks.

For an MVS environment, you can use multiple autotasks to distribute the automation workload across
multiple processors, thus taking advantage of multitasking to improve throughput. Throughput for an
automation workload can be constrained by contention for host processors or by contention for the
command list data control block (DCB), which synchronizes input and output (I/O) to the command list
data set.
For an automation workload in which the command lists have been preloaded using the LOADCL
command, no I/O is necessary to the command list data set; therefore, processor capacity is the only
constraint on automation throughput. When an n-way processor (where n is the number of processing
engines) is used with such a workload, additional autotasks beyond the number n probably do not offer a
significant throughput improvement.
This does not imply, however, that there is no value in using more than n autotasks in an n-way processor
environment. For example, it can be useful to separate the automation of critical messages from the rest
of the automation workload by assigning them to their own autotask.

Using Full-Screen Automation

Using full-screen automation, a REXX, PL/I, or C program can perform any NetView full-screen component
tasks that a NetView operator could perform using the same component and a 3270 display. Full-screen
automation is intended to be used as an automation tool. It is not intended to be used by NetView
operators to create fast paths or to simplify commands.
Using full-screen automation, an application program is able to:
• Read data from a NetView application panel
• Write data to a NetView application panel
• Execute the PF, PA, Enter, and Clear function keys on a NetView application panel
Sample CNMS1098 is included with the NetView program. This sample provides an example for coding
and using the full-screen automation function. CNMS1098 uses a BGNSESS command to logon to TSO
from the NetView program and display data from the TSO log on the NetView users panel.
Full-screen automation uses a VOST (virtual OST) task. The VOST is like an OST, except that the session
is a pipe to an owning task. Full-screen automation provides two commands (ATTACH and DETACH) to
enable procedures to invoke, control, and end full-screen applications.

14 IBM Z NetView: Tuning Guide

 The ATTACH command is used to begin a VOST and assign (queue) work to it. The DETACH command
causes a VOST started by an ATTACH to terminate.
Two commands show status of VOSTs currently active.
• LIST STATUS=VOST will show a line like the following:

TASKNAME: DSI#0001 OWNER: OPER2 ATTACHN: NLDM STATUS: ACTIVE

• TASKUTIL will show tasks with TYPE=VOST if there are VOSTs active:

TASKNAME TYPE DPR CPU-TIME N-CPU% S-CPU% MESSAGEQ STORAGE-K CMD

-------- ---- --- ----------- ------ ------ -------- --------- --------
DSI#0001 VOST 250 0.01 42.29 0.10 0 120 NLDM
DSIMONIT OPT 255 2.78 16.45 0.04 N/A 4 N/A
AAUTSKLP DST 247 0.17 14.54 0.03 0 1552 N/A
CNM01PPT PPT 255 3.29 12.52 0.03 0 164 **NONE**
DSILOG DST 254 0.15 4.31 0.01 0 19 N/A
OPER2 OST 251 1.73 3.87 0.01 0 287 **NONE**

Note:
1. DSI#xxxx is not an available choice; it is assigned by the NetView program when an ATTACH
command is issued.
2. The example shows only one VOST active at this time. TASKUTIL shows all active tasks, including
VOSTs. The TYPE column shows the type of each active task.
On average, full-screen automation uses more processor resources than manual execution of commands
or procedures. All storage allocated during the ATTACH phase of full-screen automation is released during
the DETACH step, except for the normal storage growth seen when an OST logs on. The OST storage
growth is released when the task from which the ATTACH was issued is terminated. Because of the
potential below-the-line storage growth if many VOSTs are active at once, careful consideration should
be used when determining what to automate using full-screen automation. This is true in cases where
full-screen automation is used with tasks that are not normally logged off.

Command and Message Forwarding

You can transmit commands and messages between a distributed system and a focal point by using the
RMTCMD command. This command can communicate with remote systems by using either TCP/IP or LU
6.2.

Separation of the Automation Workload from Other NetView

Workloads
In some environments, separating the automation workload from other NetView workloads can offer
advantages in automation responsiveness and availability.
You can separate workloads by:
• Running multiple copies of the NetView program
• Using z/OS Work Load Manager (WLM) enclaves to separate system and network automation workloads
within a single NetView address space

Multiple NetView Programs

You can use one copy for system automation and another copy for network automation and problem
determination.
Although separating the automation workload from other NetView workloads can offer advantages,
a disadvantage is that dividing the NetView workloads between different address spaces might not
decrease the overall NetView host processor utilization and it might increase overall NetView storage use.

Chapter 2. Tuning for Automated Operations 15

 If you plan to use multiple copies of the NetView program, consider setting the dispatching priority
of the NetView program that performs system automation above VTAM and the subsystems being
automated. The NetView program performing network automation and problem determination should
have a dispatcher priority below VTAM and critical applications.
Refer to IBM Z NetView Automation Guide for more information about running multiple copies of the
NetView program.

Single NetView Program Using WLM Enclaves

To use the Work Load Manager (WLM) for separating system and network automation workloads within
the NetView program, perform the following steps:
1. Either code or uncomment the WLM statement.
Note: If the WLM statement is not coded or uncommented, the NetView program does not use the
WLM services. A sample WLM statement is included in the CNMSTYLE member but is commented out.
For more information, refer to IBM Z NetView Installation: Configuring Additional Components.
Use the CNMSTUSR or CxxSTGEN member to code the WLM statement. See the sample statement in
the CNMSTYLE member. The name that you code for the SubSystemName keyword is the subsystem
that WLM recognizes as the specific instance of the NetView program. . For information about changing
CNMSTYLE statements, see IBM Z NetView Installation: Getting Started.
Uncommenting the statement sets the WLM subsystem name to the NetView domain ID.
2. Before you start the NetView program, create a WLM service definition. The service definition consists
of at least the following:
• A service policy
• A workload
• A service class
When separating the management of the network and system automation (SA) subtasks, the
NetView program creates z/OS enclaves to manage those two sets of subtasks so that users can
assign different performance goals to the enclaves. System automation subtasks include all those
not connected with network management.
These two types of NetView enclaves should be classified to service classes with velocity goals and
relative importance assigned. The goals should have approximately the same velocity value, but
the goal assigned to NetView system automation enclaves should be more important than the goal
assigned to any NetView network enclaves. There is no need to define a separate service class for
NetView, if existing service classes in your service definition satisfy these conditions.
If SA/390 or other system automation is used, a high velocity goal and importance could be
assigned. For non-system automation NetView subtasks, a medium velocity goal and importance
could be assigned. For example, a goal of velocity = 50 and an importance of 1 could be assigned.
For non-system automation NetView subtasks, a goal of velocity = 40 and an importance of 2 could
be assigned to give appropriate weighting to the system automation NetView subtasks.
For instructions on defining performance goals and service classes to z/OS, refer to the IBM
publication z/OS MVS Planning: Workload Management.
• Service classification rules
NetView subtasks are classified into a service class based on one or more of the following work
qualifiers:
– Subsystem type
NETV
– Subsystem instance (SI)
The NetView WLM SubSystemName statement value

16 IBM Z NetView: Tuning Guide

 – NETID (NET)
The NetView network ID
– Transaction class (TC)
The NetView subtask type: AOST, DST, HCT, MNT, NNT, OPT, OST, or PPT
Note: Virtual OSTs (VOSTs) execute in the same enclave as their owning OST.
– Transaction name (TN)
The NetView subtask module name
– User ID (UI)
The NetView subtask ID
– Logical unit name (LU)
The NetView subtask LU name
– Priority (PRI)
The NetView subtask priority
The WLM service definition must be installed on a WLM couple data set, and a WLM service policy using
that WLM service definition must be activated.
After the NetView program is started, every NetView subtask is classified into a WLM service class (based
on the customer coded WLM classification rules for the NETV subsystem type). The NetView program
joins the subtask to a WLM enclave with the same service class.
If a WLM enclave for the service class does not exist, the NetView program tells WLM to create one. When
the NetView subtask ends, the NetView program tells WLM to delete the enclave.
If no NetView subtask is joined to a particular enclave (created by the NetView program), the enclave is
deleted.

Chapter 2. Tuning for Automated Operations 17

18 IBM Z NetView: Tuning Guide
Chapter 3. Tuning for AON

AON performance considerations are described in this chapter.

Tuning Techniques
Consider the following NetView techniques when tuning AON:
• Perform NetView tuning according to the applicable information for your release of the NetView
product. You should pay particular attention to session monitor tuning and to VSAM tuning.
• In any online system, I⁄O to databases is of primary importance. Do not put the NetView logs, the
session monitor databases, the AON automation log (if used) and the AON status file all on the same
DASD, or on DASD that is highly utilized.
• Do not specify secondary extents for logs if you are using the AUTOFLIP keyword.

CNMCMD Resident Option

The resident option can potentially improve performance. For any module defined as RES=Y in the
CNMCMD initialization member, ensure that you do not override this attribute in the CNMCMDU member.
Additionally, do not override AON modules defined as RES=Y in the EZLCMENT sample.
You can make programs and REXX procedures resident with the NetView LOADCL command. This is
done automatically at AON initialization if the programs and procedures are specified with the RESIDENT
entry in the control file. A number of such entries are included in the EZLCFG01 and FKVCFG01 samples.
This list can be added to, or can be reduced, if analysis of the MAPCL output shows infrequently used
procedures.
Ensure other programs and procedures are resident as appropriate. Look in the NetView log at message
CNM493I. This message indicates that a command procedure was issued out of the automation table. All
such procedures should be resident for maximum performance. You can also use tracing to observe the
flow of AON tasks.
Note: Tracing temporarily increases system usage. Consider making resident any procedures you see that
are not already resident.

DSICLD Library
In addition to program residency, DSICLD library concatenation can also affect performance. Keep the
number of libraries to a minimum, to reduce program search time. You might want to have two user
program libraries, a small one concatenated ahead of CNMCLST and a larger one after CNMCLST. For
maximum efficiency each library should reside on a different storage directory.

Automation Table
You can use the BEGIN/END feature to group all of your messages together in short sections. This cuts
automation table processing time considerably. Change the order of the IF THEN statements in the
automation table, so that the statements are ordered by frequency, which places the most frequently
processed message at the beginning. You can determine this by issuing the AUTOCNT command. For
more information about the AUTOCNT command, see the IBM Z NetView Command Reference Volume 1
(A-N) or the NetView online help.
Code early outs in the automation table. That is, if you are aware of messages that are being passed to
the NetView program and are not driving programs out of the automation table, code IF THEN statements
to exit near the beginning of your table. For example, if you know that there are no message triggers for

© Copyright IBM Corp. 1997, 2022 19

 IEF messages (and these messages are not being stopped by MPF list with AUTO=N), you can code the
following statement at the beginning of your automation table:

IF TEXT(1)='IEF'. THEN DISPLAY(N)

Ordinarily, messages such as these require an entry in the MPF list specifying AUTO=N, so that the
NetView product does not have to process the messages at all. If you are using another NetView program
for console automation, these messages must come over the subsystem interface (SSI). In these cases,
early outs are necessary in the AON automation table.
To further tune message processing, follow all of the preceding suggestions and then at the end of your
automation table, code an entry that traps on all other messages (IF MSGID=.) and runs a program that
writes that message to a file. Then, do whatever possible to omit them from processing.

DDF Tree and Panel

Ensure that EZLPNLS contains all DDF panels that you want to access. This slows down DDF initialization
just a bit, but speeds up operator access and DDF update later.
Use the dynamic updating of resources by type (as included with AON) as much as possible. Avoid hard
coding resources onto panels and avoid having them as unique entries in EZLTREE. Keep EZLTREE as
small as possible.

Node Automation
AON does not know when a resource goes away suddenly, or if the outage was intentional, unless an
operator has specified it as inactive. For example, You might have specified a switched PU on a token
ring or an X.25 line is to be dropped after business hours. If your network has many such devices,
use RECOVERY control file entries with the NOAUTO parameter to forestall pointless recovery attempts
during off hours. You can code these statements generically with wildcard names, if you have a naming
convention, to decrease the number of statements required.

Operations
If you must recycle the AON NetView program during peak times of the day, do not code DDFREFRESH=Y
on the ENVIRON SETUP control file entry. This means that DDF does not know about resources that are
already down, but it is updated when they are running again. You can add them later by a restricted
NETSTAT call, or manually with DDFADD.
NETSTAT uses considerable cycles if TYPE=PHYSICAL is required. To manually add an AON resource to
DDF, you need to know the status you want it to show, which is determined by priority. The priority
value is found on the DDF entry in the control file (PR=nnn). For example, the priority for inactive is 150
(DDF INA*,PR=150,CLEAR=Y). When you add the resource to DDF with an inactive status, you want to do
so in such a way that DDF deletes it automatically when the resource comes active. This requires that
you specify on the DDFADD the same basic information that AON does. Specify the information in the
following format:

DDFADD sysname.resource_name(resource_type),
RV=resource_name,IN=/resource_name/,DA='some text',
PR=nnn

The DA field must contain some message text but it need not be the same text AON uses.
Use DBMAINT sparingly online. It is important to clean up and reorganize your databases to maximize
performance, but DBMAINT uses a lot of cycles and should not be used during peak periods. Set a timer
to run it during off hours.
Keep the number of operators who are notification operators to a minimum to save CPU cycles. Many
operators often prefer DDF to the messages.

20 IBM Z NetView: Tuning Guide

TCP/IP Support for AON
To reduce monitoring overhead, avoid monitoring noncritical resources. Define low monitor intervals only
for the highly critical resources and define greater monitor intervals for resources which might be less
critical.
Monitor the TASKUTIL message queues and DISPPI buffer queues and increase the number of TSO
servers and AUTTCP tasks and AUTMSG tasks as required to reduce the backlog of queues.

Chapter 3. Tuning for AON 21

22 IBM Z NetView: Tuning Guide
Chapter 4. Tuning for Command Procedures

A command list is a list of commands and statements designed to perform a specific function for the user.
For the NetView program, command lists can be written using the NetView command list language or
REXX. In the NetView program, a command processor is a module written in a high-level language (HLL)
or assembler language and is invoked as a command. A command procedure is either a command list or a
command processor written in PL/I or C.
For more information on command lists and command processors, see the following publications:
• IBM Z NetView Customization Guide
• IBM Z NetView Programming: Assembler
• IBM Z NetView Programming: PL/I and C
• IBM Z NetView Programming: REXX and the NetView Command List Language

Tuning Techniques
Following are the major tuning techniques for command procedures, arranged in order of expected
improvement on performance; the most important tuning considerations are listed first. These are
described in detail in this chapter.
1. Preload frequently used command lists using the LOADCL command. Consider invoking a command
list to issue LOADCL, loading the most frequently used command lists last. See “Command Lists” on
page 24.
2. For systems using PL/I command processors, define preinitialized environments for your frequently
run PL/I programs. You can use the HLLENV LIST STATS command to help you decide how many
preinitialized environments you need. See “Running High-level Language Programs in a Preinitialized
Environment” on page 32.
3. For systems using PL/I or C command processors, minimize or eliminate the I/O needed to find
and read the runtime routines during command processor execution. See “Command Processors” on
page 31.
4. Preload frequently used command processors by not specifying RES=N on their CMDDEF statements
in the CNMCMDU initialization member. See “Command Processors” on page 31.
5. Consider using the REXX/370 compiler to compile REXX command lists. Compiled REXX command
lists are more efficient than interpreted REXX command lists. See “Compiled REXX/370 Command
Lists” on page 26.
6. Avoid using nested command lists for subroutines where possible. See “Subroutines” on page 25.
7. For systems using REXX command lists, group frequently used external functions and subroutines in
REXX function packages. You can also modify the sample file CNMSJM11 to minimize the search time
for NetView system functions. See “REXX Function Packages” on page 27.
8. Use task global variables instead of common global variables where possible. The processing
demands of task global variables are smaller than the processing demands of common global
variables. See “Global Variables” on page 34.
9. If you expect to use more than 200 task or common global variables, specify the expected number
of variables in the NetView constants module DSICTMOD. You can use the QRYGLOBL command
to display the expected number of variables coded in DSICTMOD, as well as the actual number of
variables found. See “Global Variables” on page 34.
10. Consider using or adapting the AUTODROP command list (CNMS8003) to manage preloaded
command lists. See “Managing Command Lists with AUTODROP” on page 25.

© Copyright IBM Corp. 1997, 2022 23

Command Lists
Command lists that are written using the NetView command list language or REXX are interpreted. That is,
statements in the command list are translated and executed one at a time.
Before being interpreted, the entire command list is read into storage (unless you used LOADCL).
However, trailing blanks are removed from command lists to save storage. Each time a command list is
invoked, it is read from the command list data set (DSICLD). When users modify command lists, changes
are put into effect immediately. Nested command lists are read into storage as they are invoked.

Preloading Command Lists

You can preload frequently used command lists into storage with the LOADCL command. Preloading
command lists saves time. When a preloaded command list is invoked, it executes immediately. There
is no I/O delay. Preloaded command lists are placed above the 16 MB line in MVS. If command lists are
preloaded, every operator uses the same copy. If command lists are not preloaded, multiple copies of the
same command list can be loaded when invoked by different operators. This can cause storage problems
depending on the size of the command list and the frequency of invocations.
The DROPCL command removes a command list that was previously loaded into storage by the LOADCL
command. The MAPCL command displays the command lists that have been preloaded, the date and time
each was loaded into storage, and the number of invocations for each command list since it was loaded.
If you plan to preload several command lists using the LOADCL command, consider writing a command
list to issue the LOADCL command invocations. Order the LOADCL command invocations so that the most
frequently used command lists are loaded last.
You can use the detail report produced by the AUTOCNT command to select command lists for
preloading. Look for automation table statements with a high match count that execute commands.
Preload command lists that are executed by these statements.
You can also use the PIPE stage INSTORE to add, delete, or replace frequently used command lists into a
pool of storage. The in-storage members are read from storage instead of disk by DSIDKS disk services or
any NetView process based on DSIDKS, such as BROWSE or the < stage.
Using the STRIP stage with PIPE INSTORE enables you to strip trailing blanks before loading, thus
reducing storage demand. When the CRYPTO parameter is used with PIPE INSTORE, the member is
encrypted when it is loaded into memory. This should be used only when the security of the member is
important, because additional processing is required to encrypt the data when this stage is executed and
decrypt the data each time the member is read.
The INSTORE stage is used by the REXX CLIST CNME1054 (MEMSTORE) to automatically load frequently
used members. MEMSTORE will use the LIST MEMSTAT command to monitor members with elevated
disk usage, load these members into memory, or unload members that have relatively low usage from
memory.
Using the two parameters for MEMSTORE allows you to control the amount of storage allocated to
in-storage members, which resides above the 16 MB line in MVS, and the monitoring interval used to
obtain member usage information.
The amount of processing for MEMSTORE is directly related to the monitoring interval. LOADCL might
run a CLIST slightly faster, but storage usage can be optimized better with MEMSTORE. Also, MEMSTORE
monitors and adjusts CLIST loading based upon usage without user intervention.
Note that the NetView program authority checks all REXX procedures and command lists before any are
loaded. If REXX procedures or command lists do not show as being loaded (using the MAPCL command),
users must verify that the operator attempting to load them has the proper authority. See the IBM Z
NetView Security Reference for more information.

24 IBM Z NetView: Tuning Guide

Managing Command Lists with AUTODROP
A sample command list, AUTODROP (CNMS8003), provided by the NetView program can help you
manage the number of command lists that have been preloaded into storage using the LOADCL command.
AUTODROP uses the MAPCL and DROPCL commands to conditionally drop command lists from storage.
AUTODROP uses the following parameters:
Uses
Specifies the minimum number of times the command list must be used to remain in storage. The
default is 0.
Days
Specifies the minimum number of days the command list must be loaded to be dropped from storage.
The default is 0.
Hours
Specifies the minimum number of hours the command list must be loaded to be dropped from
storage. The default is 1.

Example: Using the AUTODROP command

The AUTODROP command drops any preloaded command lists with a use count of five or less that has
been loaded for three hours or more, as shown in the following example:

AUTODROP 5 0 3

A command is not dropped from storage unless both the minimum time specified by the days and hours
parameters has elapsed and the command list has been invoked the minimum number of times necessary
(or less) since the command list was loaded.

Determining Command Lists to Drop

You can use the AUTODROP command list to manage commands preloaded with LOADCL. Use the
following steps to determine the subset of the most frequently used command lists:
1. Preload a large subset of your command lists and then issue AUTODROP from a timer to remove
infrequently used command lists.
2. Drop and reload the command lists that are to stay resident to reset their load times and use counts.
Note: Reload the command lists in ascending order of frequency, with the most heavily used ones last.
3. Preload another large subset and repeat the procedure until all command lists have been processed.
If the storage requirement for the most frequently used command lists is too high, consider making the
AUTODROP criteria more stringent, with a higher use count or a shorter time interval.

Subroutines
Instead of using nested command lists for subroutines, consider calling the subroutine, as shown in the
following examples:
• for the NetView command list language:

&RETURN = -RI
&GOTO -SUB1
-RI
⋮
&EXIT
-SUB1
⋮
&GOTO &RETURN

Chapter 4. Tuning for Command Procedures 25

 • for REXX:

CALL SUB1
⋮
EXIT
SUB1: PROCEDURE
⋮
RETURN

If you use nested REXX command lists, do not use the CALL statement to invoke them. Instead, place
the command list name in single quotation marks. The CALL statement causes the system libraries to be
searched before the NetView command list library. When the command list name is in single quotation
marks, the command list name is passed to the NetView program, and only the NetView library is
searched.

REXX Command Lists

Most command lists written in REXX perform more efficiently than command lists written in the NetView
command list language. This is true for command lists that perform mathematical functions or string
manipulation. However, processing global variables in REXX command lists can require more processing
time than NetView command list language command lists. Note the following considerations:
• When global variables are identified in NetView command lists by &TGLOBAL and &CGLOBAL
statements, all subsequent assignments to the variables cause updates to the global variables. In
REXX, the programmer has explicit control over access to global variables using the NetView GLOBALV
command.
• When the NetView program processes a GLOBALV command issued from within a REXX command list,
it accesses both the global variable and the REXX local variable corresponding to the global variable.
Accessing the REXX local variable during NetView GLOBALV command processing requires processing
time not needed for NetView command lists.
• You can minimize the use of the GLOBALV command in REXX to optimize the NetView global variable
processing. With REXX command lists, the REXX local variable corresponding to the global variable can
be manipulated many times without updating the global variable.

REXX local variables are accessed while processing other NetView commands, such as GETMTYPE,
GETMSIZE, and PARSEL2R. These commands pass information back to the REXX command list by
updating the REXX local variables. These NetView commands require more processing time when invoked
from REXX command lists than when invoked from NetView command lists.

How to Compile REXX Procedures

You can compile AON REXX programs for improved performance. The REXX programs in the RESIDENT
entries of the AON control file are good choices for compiling.
Note: Module level tracing is not available for compiled REXX programs, although entry and exit tracing is
available. You might need to reinstate the interpreted versions of the REXX programs to obtain traces for
problem diagnosis.
For more information about using the REXX compiler, refer to the IBM REXX/370 Compiler and Library
User's Guide and Reference.

Compiled REXX/370 Command Lists

You can use the REXX/370 compiler to compile REXX command lists written for execution on the NetView
program. Compiled REXX command lists are more efficient than interpreted REXX command lists. When
compiling REXX command lists, specify the compiled EXEC (CEXEC) output format. You can do the
compilation on a separate MVS system with the REXX/370 compiler installed.

26 IBM Z NetView: Tuning Guide

 Storage Considerations
The size of a compiled REXX command list often exceeds the size of the source command list. The
CONDENSE compiler option enables you to significantly reduce the size of the CEXEC type output. Use the
CONDENSE compiler option to:
• Reduce the amount of disk space required by compiled REXX command lists.
• Reduce the amount of virtual storage required by compiled REXX command lists preloaded with the
LOADCL command.
• Reduce the amount of I/O activity required to load compiled REXX command lists.
When a condensed compiled REXX command list is invoked, the command list is automatically
uncondensed. A condensed compiled REXX command list requires more storage while it is running, for
the following reasons:
• During the uncondense operation, an additional 128 KB of storage is required.
• While a condensed compiled REXX command list is running, both the condensed and uncondensed copy
exist in storage.
Additional CPU time is required to uncondense the compiled REXX command list. Otherwise, the
performance characteristics of a condensed compiled command list are the same as those of an
uncondensed compiled command list.

Performance Considerations
The performance improvements that you can expect when you run a compiled REXX command list
depend on the type of command list. A program that performs many arithmetic operations shows
the greatest improvement. A program that mainly issues commands (such as NetView, VTAM, or MVS
commands) shows limited improvement, because the compiler cannot decrease the time taken in
processing the commands.

REXX Function Packages

REXX function packages are groups of external functions and subroutines that are packaged together.
When the REXX language processor processes a function call or a call to a subroutine, the language
processor searches the function packages before searching load libraries. Grouping frequently used
external functions and subroutines in a function package allows faster access to the functions and
subroutines, resulting in better performance.
Both the NetView program and REXX interpreter can have user, local, and system function packages.
The functions for REXX that are supplied with the NetView product are in the NetView system function
package. When a function is called, the function package search order is:
1. NetView user
2. REXX interpreter user
3. NetView local
4. REXX interpreter local
5. NetView system
6. REXX interpreter system

You can improve the performance of NetView system functions by making the NetView system function
package higher in the search order so that it can be found sooner. Take the following steps to modify the
function package section of the NetView sample file CNMSJM11. The labels in the following steps refer to
the examples in Figure 4 on page 28, Figure 5 on page 29, and Figure 6 on page 30.
1. Move PACKTB_NAME from the bottom of the list to either replace or follow PACKTB_USER_NAME
(Figure 4 on page 28 shows the initial order), depending on if you have defined your own user
function package.

Chapter 4. Tuning for Command Procedures 27

 2. Adjust the NetView function package table header counters.
• If you replace PACKTB_USER_NAME with PACKTB_NAME, change PACKTB_SYSTEM_TOTAL and
PACKTB_SYSTEM_USED from "1" to "0" (see Figure 5 on page 29).
• If you have PACKTB_NAME following PACKTB_USER_NAME, change PACKTB_USER_TOTAL and
PACKTB_USER_USED from "1" to "2", and PACKTB_SYSTEM_TOTAL and PACKTB_SYSTEM_USED
from "1" to "0" (see Figure 6 on page 30).
3. Assemble and link edit the sample file.
The following figures illustrate how you can change the search order of function packages. The first figure
represents the sample file as it is included with the NetView program. The second and third figures show
how to modify the file if you did not define a user function package, or if you did define a user function
package, respectively.

*/********************************************************************/
*/* */
*/* PACKTB_HEADER - NetView REXX Function Package Table Header*/
*/* */
*/********************************************************************/
SPACE 3
***********************************************************************
* NOTE: Do NOT change any of the following address constant fields.
* The total number and number of used PACKTB entries may be
* changed as needed.
***********************************************************************
SPACE
PACKTB_HEADER DS 0C REXX Function Package Table HDR
PACKTB_USER_FIRST DC A(PACKTB_USER_ENTRY) Address of first USER
* PACKTB entry
PACKTB_USER_TOTAL DC F'1' Total number of USER PACKTB
* entries
PACKTB_USER_USED DC F'1' Number of used USER PACKTB
* entries
PACKTB_LOCAL_FIRST DC A(PACKTB_LOCAL_ENTRY) Address of first LOCAL
* PACKTB entry
PACKTB_LOCAL_TOTAL DC F'1' Total number of LOCAL PACKTB
* entries
PACKTB_LOCAL_USED DC F'1' Number of used LOCAL PACKTB
* entries
PACKTB_SYSTEM_FIRST DC A(PACKTB_SYSTEM) Address of first SYSTEM
* PACKTB entry
PACKTB_SYSTEM_TOTAL DC F'1' Total number of SYSTEM PACKTB
* entries
PACKTB_SYSTEM_USED DC F'1' Number of used SYSTEM PACKTB
* entries
PACKTB_LENGTH DC F'8' Length of each PACKTB entry
PACKTB_FFFF DC XL8'FFFFFFFFFFFFFFFF' End marker
SPACE 3
*/********************************************************************/
*/* */
*/* PACKTB_ENTRY - NetView REXX Function Package Table Entries*/
*/* */
*/********************************************************************/
SPACE 2
***********************************************************************
* NOTE: Do NOT change the NetView SYSTEM Function Package name. The
* LOCAL and USER Function Package names may be changed and
* additional SYSTEM Function Package Table entries may be made.
***********************************************************************
SPACE
PACKTB_USER_ENTRY DS 0C REXX USER Function Package Table
* Entry
PACKTB_USER_NAME DC CL8'DSIRXUFP' Name of USER Function Package
PACKTB_LOCAL_ENTRY DS 0C REXX LOCAL Function Package
* Table Entry
PACKTB_LOCAL_NAME DC CL8'DSIRXLFP' Name of LOCAL Function Package
PACKTB_SYSTEM DS 0C REXX SYSTEM Function Package
* Table Entry
PACKTB_NAME DC CL8'DSIRXFPG' Name of SYSTEM Function Package
SPACE 3
END DSIRXPRM End of DSIRXPRM module

Figure 4. CNMSJM11 before Modification to Switch Function Package Search Order

28 IBM Z NetView: Tuning Guide

 */********************************************************************/
*/* */
*/* PACKTB_HEADER - NetView REXX Function Package Table Header*/
*/* */
*/********************************************************************/
SPACE 3
***********************************************************************
* NOTE: Do NOT change any of the following address constant fields.
* The total number and number of used PACKTB entries may be
* changed as needed.
***********************************************************************
SPACE
PACKTB_HEADER DS 0C REXX Function Package Table HDR
PACKTB_USER_FIRST DC A(PACKTB_USER_ENTRY) Address of first USER
* PACKTB entry
PACKTB_USER_TOTAL DC F'1' Total number of USER PACKTB
* entries
PACKTB_USER_USED DC F'1' Number of used USER PACKTB
* entries
PACKTB_LOCAL_FIRST DC A(PACKTB_LOCAL_ENTRY) Address of first LOCAL
* PACKTB entry
PACKTB_LOCAL_TOTAL DC F'1' Total number of LOCAL PACKTB
* entries
PACKTB_LOCAL_USED DC F'1' Number of used LOCAL PACKTB
* entries
PACKTB_SYSTEM_FIRST DC A(PACKTB_SYSTEM) Address of first SYSTEM
* PACKTB entry
PACKTB_SYSTEM_TOTAL DC F'0' Total number of SYSTEM PACKTB
* entries
PACKTB_SYSTEM_USED DC F'0' Number of used SYSTEM PACKTB
* entries
PACKTB_LENGTH DC F'8' Length of each PACKTB entry
PACKTB_FFFF DC XL8'FFFFFFFFFFFFFFFF' End marker
SPACE 3
*/********************************************************************/
*/* */
*/* PACKTB_ENTRY - NetView REXX Function Package Table Entries*/
*/* */
*/********************************************************************/
SPACE 2
***********************************************************************
* NOTE: Do NOT change the NetView SYSTEM Function Package name. The
* LOCAL and USER Function Package names may be changed and
* additional SYSTEM Function Package Table entries may be made.
***********************************************************************
SPACE
PACKTB_USER_ENTRY DS 0C REXX USER Function Package Table
* Entry
PACKTB_NAME DC CL8'DSIRXFPG' Name of SYSTEM Function Package
PACKTB_LOCAL_ENTRY DS 0C REXX LOCAL Function Package
* Table Entry
PACKTB_LOCAL_NAME DC CL8'DSIRXLFP' Name of LOCAL Function Package
PACKTB_SYSTEM DS 0C REXX SYSTEM Function Package
* Table Entry
SPACE 3
END DSIRXPRM End of DSIRXPRM module

Figure 5. CNMSJM11 After Modification with No User Function Package Defined

Chapter 4. Tuning for Command Procedures 29

 */********************************************************************/
*/* */
*/* PACKTB_HEADER - NetView REXX Function Package Table Header*/
*/* */
*/********************************************************************/
SPACE 3
***********************************************************************
* NOTE: Do NOT change any of the following address constant fields.
* The total number and number of used PACKTB entries may be
* changed as needed.
***********************************************************************
SPACE
PACKTB_HEADER DS 0C REXX Function Package Table HDR
PACKTB_USER_FIRST DC A(PACKTB_USER_ENTRY) Address of first USER
* PACKTB entry
PACKTB_USER_TOTAL DC F'2' Total number of USER PACKTB
* entries
PACKTB_USER_USED DC F'2' Number of used USER PACKTB
* entries
PACKTB_LOCAL_FIRST DC A(PACKTB_LOCAL_ENTRY) Address of first LOCAL
* PACKTB entry
PACKTB_LOCAL_TOTAL DC F'1' Total number of LOCAL PACKTB
* entries
PACKTB_LOCAL_USED DC F'1' Number of used LOCAL PACKTB
* entries
PACKTB_SYSTEM_FIRST DC A(PACKTB_SYSTEM) Address of first SYSTEM
* PACKTB entry
PACKTB_SYSTEM_TOTAL DC F'0' Total number of SYSTEM PACKTB
* entries
PACKTB_SYSTEM_USED DC F'0' Number of used SYSTEM PACKTB
* entries
PACKTB_LENGTH DC F'8' Length of each PACKTB entry
PACKTB_FFFF DC XL8'FFFFFFFFFFFFFFFF' End marker
SPACE 3
*/********************************************************************/
*/* */
*/* PACKTB_ENTRY - NetView REXX Function Package Table Entries*/
*/* */
*/********************************************************************/
SPACE 2
***********************************************************************
* NOTE: Do NOT change the NetView SYSTEM Function Package name. The
* LOCAL and USER Function Package names may be changed and
* additional SYSTEM Function Package Table entries may be made.
***********************************************************************
SPACE
PACKTB_USER_ENTRY DS 0C REXX USER Function Package Table
* Entry
PACKTB_USER_NAME DC CL8'DSIRXUFP' Name of USER Function Package
PACKTB_NAME DC CL8'DSIRXFPG' Name of SYSTEM Function Package
PACKTB_LOCAL_ENTRY DS 0C REXX LOCAL Function Package
* Table Entry
PACKTB_LOCAL_NAME DC CL8'DSIRXLFP' Name of LOCAL Function Package
PACKTB_SYSTEM DS 0C REXX SYSTEM Function Package
* Table Entry
SPACE 3
END DSIRXPRM End of DSIRXPRM module

Figure 6. CNMSJM11 After Modification with a User Function Package Defined

Refer to IBM Z NetView Programming: Assembler for information about adding functions to the NetView
function packages.

Tuning REXX Environments

Before the REXX language processor can process a REXX command list, a REXX language processor
environment must exist. A REXX language processor environment is the environment in which the REXX
language processor interprets or processes the command list. This environment defines characteristics
relating to how the command list is processed and how the language processor accesses system services.
When a REXX command list is executed in the NetView program, the REXX interpreter sets up a language
processor environment for the NetView program. When the command list ends, this unique environment
can be held for reuse by the same task. The NetView program retains these REXX environments to
improve REXX environment initialization performance.

30 IBM Z NetView: Tuning Guide

 Three operands of the DEFAULTS and OVERRIDE commands affect how REXX environments are handled:
• REXXENV is the maximum number of REXX environments retained for a task. If you set this number
to zero, the NetView program does not save any REXX environment. This adversely affects the
performance of all NetView REXX command lists.
• REXXSTOR is the amount of storage allocated when a REXX environment is initialized. The default
amount is determined by the REXX interpreter, and is sufficient for REXX command lists with up to
six levels of nested invocations. If you set this number to zero, storage is acquired as needed, but
performance is degraded for the first command list using a given REXX environment.
• REXXSLMT is the maximum amount of storage that a REXX environment is allowed to accumulate
before being ended, after its current use is completed. Storage associated with a REXX environment
can increase, depending on the needs of the REXX command lists that have been interpreted under the
environment. Because each REXX command list can have different storage needs, REXX environments
can grow to meet the storage needs of the most demanding REXX command list. If the storage for an
environment grows to exceed the amount of storage specified by REXXSLMT, the environment is freed
when the REXX command list finishes executing.
Note: The values for REXXENV, REXXSLMT, and REXXSTOR do not apply to Data REXX environments.
REXXSTOR is the lower limit of storage in each environment. REXXSLMT is an upper threshold to prevent
environments that grow very large from being allocated and not freed until the owning task terminates.
The following are tuning considerations for the REXXENV, REXXSTOR and REXXSLMT operands of the
DEFAULTS and OVERRIDE commands:
• For most situations, a small REXXENV value (1 or 2) is sufficient, because most operators will have only
one or two REXX command lists active at once.
• Consider setting REXXSTOR to zero, because it affects only the first REXX command list interpreted
under a REXX environment. A value of zero minimizes the potential for storage waste.
• Set a value for REXXSLMT so that REXX environment storage does not grow indefinitely.
• If you want to minimize storage at the expense of extra processing time, set both REXXENV and
REXXSTOR to zero. With these settings, storage for REXX environments is allocated only when needed,
and is freed as soon as possible.
• Use the TASKUTIL command to monitor the amount of queued storage allocated for different tasks.
REXX environments are allocated as queued storage under operator station tasks (OSTs) or autotasks.
The environments are part of the storage total for the task shown in output from the TASKUTIL
command. If you notice a significant storage growth for a task that is executing many REXX command
lists, consider adjusting the values for REXXENV, REXXSTOR, and REXXSLMT.
Note:
1. An entry (2 slots) in the REXX anchor table, IRXANCHR, is required for each non-nested REXX
command list to run.
2. If a REXX command list is called from another REXX command list, a new environment is not required.
The nested command list uses the environment of the primary command list.

If you want information about... Refer to...

The DEFAULTS and OVERRIDE commands IBM Z NetView Command Reference Volume 1 (A-N)
Nesting REXX command lists from the assembler, IBM Z NetView Programming: REXX and the NetView
PL/I, or C languages Command List Language
Estimating and setting the number of REXX IBM Z NetView Installation: Configuring Additional
environments Components

Command Processors
Each command processor must have a CMDDEF statement in the CNMCMD initialization member. Do not
specify RES=N on the CMDDEF statements for frequently used command processors. If you do not specify

Chapter 4. Tuning for Command Procedures 31

 RES=N, the modules are made resident, eliminating the I/O needed to load them every time they are
called.
You can use the AUTOCNT command to identify command procedures that are executed frequently from
the automation table. See “AUTOCNT Command” on page 9 for more information.
If you are using the generic automation receiver (NVAUTO), modify the CMDDEF statement for DSINVGRP
to make the command processor resident.

Command Processors Written in a High-Level Language

Command processors that are written in an HLL are compiled rather than interpreted. In terms
of performance, compiled command procedures are usually processed much faster than interpreted
command procedures. If you are using PL/I programs, consider running them in a preinitialized
environment. See “Running High-level Language Programs in a Preinitialized Environment” on page 32
for more information.
You can improve initialization overhead for command processors that do not use preinitialized
environments by avoiding I/O in searching for and fetching the HLL runtime libraries. Consider the
following:
• If your NetView startup job control language (JCL) has a STEPLIB DD statement, eliminate it as
discussed in “STEPLIB DD Statements” on page 107. This eliminates the I/O needed to search the
STEPLIB for each LOAD, LINK, or XCTL system macro executed. If a STEPLIB DD statement exists, I/O is
required to search for the HLL runtime libraries.
• If you use command processors written in PL⁄I or C, you can preload the PL⁄I and C runtime libraries by
coding the following CMDDEF statements in CNMCMDU:

CMDDEF.IBMBLIIA.MOD=IBMBLIIA
CMDDEF.IBMBLIIA.TYPE=R
CMDDEF.IBMBLIIA.RES=Y

CMDDEF.EDCXV.MOD=EDCXV
CMDDEF.EDCXV.TYPE=R
CMDDEF.EDCXV.RES=Y

CMDDEF.EDCX24.MOD=EDCX24
CMDDEF.EDCX24.TYPE=R
CMDDEF.EDCX24.RES=Y

Preloading the HLL runtime libraries eliminates I⁄O for the directory search and the fetch for runtime
libraries.
See the appropriate MVS publication for more information about improving production library
performance.
Set the STACK and HEAP sizes carefully for your HLL command processors. For best performance, the
initial stack allocation should be large enough to satisfy all requests for stack storage. Likewise, the initial
heap segment should be large enough to satisfy all requests for heap storage. Use the facilities provided
by the HLL for tuning the STACK and HEAP sizes.

Running High-level Language Programs in a Preinitialized Environment

Before a command processor written in a high-level language (HLL) can be executed, a runtime
environment (or execution environment) must exist. A runtime environment is a set of resources that
are used to support the execution of a program. Preinitialization allows the runtime environment to
be initialized once and used for multiple program executions. Without preinitialization, the runtime
environment is created and terminated for each program execution.
You can use the HLLENV command to define the number of preinitialized environments that are allocated
by the NetView program. The HLLENV command enables you to specify two pools of preinitialized
environments using the REGENVS and CRITENVS keywords.

32 IBM Z NetView: Tuning Guide

• Regular environments (REGENVS) are allocated when the HLLENV command is issued, and are available
to any PL/I or C command processor that is capable of using preinitialized environments.
– If you want a PL/I command processor to use a preinitialized environment, link the command
processor with the DSIEXAPP interface module instead of the DSIEXANP interface module.
– If you want a C command processor to use a preinitialized environment, link the command processor
with the DSIEXAPC interface module instead of the DSIEXANC interface module.
• Critical environments (CRITENVS) are available only to PL/I and C command processors capable
of using preinitialized environments that also have bit 4 in the HLLOPTS variable set to 1. Critical
environments are not allocated when the HLLENV command is issued, but are allocated as needed.
Allocation of environments from the two pools works as follows:
• For command processors that are capable of using preinitialized environments but that do not have bit 4
set in HLLOPTS:
– If an environment is available from the regular pool, it is used.
– Otherwise, a new environment is created for this command processor, and the environment is
terminated when the command processor completes execution.
• For command processors that are capable of using preinitialized environments and that have bit 4 set in
HLLOPTS (critical command processors):
– If an environment is available from the regular pool, it is used.
– If an environment is not available from the regular pool but is available from the critical pool, it is
used.
– Otherwise, a new environment is created for this command processor, and the environment is
terminated when the command processor completes execution. If the number of environments
in the critical pool is less than the CRITENVS value specified with the HLLENV command, a new
environment is created and added to the critical pool. This makes it more likely that a preinitialized
environment will be available the next time a critical command processor is executed.
Preinitialized language environment support in the IBM Z NetView program can help improve the
performance of NetView high-level language (HLL) programs. Users can request language environment
initializations from IBM z/OS Language Environment® through the IBM Z NetView program, keep these
language environments, and then assign these environments to NetView HLL programs when needed. The
initialization of a language environment occurs only once, rather than every time an HLL program is called.
Two preinitialized language environment characteristics that you can control are the size of the primary
stack frame extent (stack extent), which houses program saveareas and autodata, and the size of the
primary heap extent (HLL program-obtained storage). These controls are provided by the PSTACK and
PHEAP keywords of the HLLENV command. The default value for these keywords is 4K (4096) bytes. You
can modify these values for performance considerations. The maximum value for either PSTACK or PHEAP
is 128K bytes.
Note: The default 4KB byte values are somewhat small for many NetView HLL programs. When a stack
frame extent or heap extent fills, Language Environment creates another extent, which is freed when it
is no longer used. This spends processor cycles on something other than the HLL program, reducing the
benefit gained by preinitializing the language environment.
The secondary stack and heap extent sizes each have a hard coded value of 4K bytes.
For more information about preinitialized environments and the HLLENV command, see IBM Z NetView
Programming: PL/I and C.

Suggestions for Using Preinitialized Environments

• Review your PL/I and C command processors and decide which ones are good candidates for using
preinitialized environments. Good candidates include installation exits and frequently used command
processors, such as those issued from the NetView automation table. Some restrictions apply to
command processors using preinitialized environments. Refer to the IBM Z NetView Programming: PL/I
and C for more details.

Chapter 4. Tuning for Command Procedures 33

 • Modify the CNMSTUSR or CxxSTGEN member to issue the HLLENV command to set up the environment
pools.
• Use the HLLENV LIST STATS command to monitor usage of the regular and critical environment pools.
Figure 7 on page 34 contains sample output for the preinitialized PL/I environment and Figure 8 on
page 34 contains sample output for the preinitialized C environment.

BNH040I IBMADPLI PREINITIALIZED ENVIRONMENT STATISTICS

BNH041I STATISTICS RESET AT: 03/30/13 16:44:04
BNH042I PSTACK: 131072 PHEAP: 131072 DEFAULT: NOTPREINIT
BNH043I NUMBER REQUESTED. REGENVS: 4 CRITENVS: 2
BNH044I ALLOCATED. REGENVS: 4 CRITENVS: 1
BNH045I IN USE. REGENVS: 4 CRITENVS: 1
BNH046I MOST NEEDED. REGENVS: 4 CRITENVS: 1
BNH047I TIMES USED. REGENVS: 4 CRITENVS: 1
BNH048I TIMES UNAVAILABLE. REGENVS: 0 CRITENVS: 1
BNH049I AVERAGE NEEDED. REGENVS: 2.50 CRITENVS: 1.00

Figure 7. Sample Output of the HLLENV, TYPE=IBMADPLI,LIST,STATS,RESET Command

BNH040I IBMADC PREINITIALIZED ENVIRONMENT STATISTICS

BNH041I STATISTICS RESET AT: 03/30/13 16:44:04
BNH042I PSTACK: 131072 PHEAP: 131072 DEFAULT: NOTPREINIT
BNH043I NUMBER REQUESTED. REGENVS: 4 CRITENVS: 2
BNH044I ALLOCATED. REGENVS: 4 CRITENVS: 1
BNH045I IN USE. REGENVS: 4 CRITENVS: 1
BNH046I MOST NEEDED. REGENVS: 4 CRITENVS: 1
BNH047I TIMES USED. REGENVS: 4 CRITENVS: 1
BNH048I TIMES UNAVAILABLE. REGENVS: 0 CRITENVS: 1
BNH049I AVERAGE NEEDED. REGENVS: 2.50 CRITENVS: 1.00

Figure 8. Sample Output of the HLLENV LIST STATUS TYPE=IBMADC

The goal of tuning the regular and critical pool allocations is to minimize the number of times that
preinitialized environments are not available when they are needed.
• Use of preinitialized environments is a processing versus storage trade-off. For the additional storage
requirement of preinitialized environments, you have a reduction in system processing. You can monitor
NetView storage usage with the RESOURCE command. Some environment storage (stack storage) is
allocated below the 16 MB line. Therefore, even if you do not have a 31 bit storage constraint, you
should monitor 24 bit storage usage.
• The storage associated with each preinitialized environment can grow over time, depending on the
needs of the PL/I or C command processors using them. Because the storage requirements for different
PL/I or C command processors vary, all the preinitialized environments eventually grow over time to
meet the needs of the most demanding PL/I or C command processor. To limit this growth, periodically
change the number of allocated environments to zero with the HLLENV command, allowing them to be
available. Then reset the requested number of environments with the HLLENV command to have them
reallocated.
For more information on preinitialized environments and the HLLENV command, refer to the IBM Z
NetView Programming: PL/I and C.

Global Variables
The following are classes of NetView global variables:
• Task global variables are accessible to any command procedure running under the task, as well as from
the automation table using the DSITGLOB automation table function (ATF). Task global variables can be
set to null, but when the variables are created, some storage is allocated for the variables until the task
ends.
• Common global variables are accessible by any command procedure running under the NetView
program from the automation table using the DSICGLOB ATF. Common global variables can be set
to null, but when the variables are created, some storage is allocated for the variables until the NetView
program ends. Common global variables can be updated directly from any task. Access to the common
global variables is serialized to provide data integrity using the system enqueue and dequeue facility.

34 IBM Z NetView: Tuning Guide

 Accessing task global variables is faster than accessing common global variables, because task global
variables do not require overhead of the system enqueue and dequeue facility. Keep this in mind when
deciding whether to use task or common global variables for an application.

Enhancing Performance
If you are writing an assembler command processor that updates multiple global variables, consider using
the NUMVARS option of the DSIVARS macro. This updates multiple global variables with a single macro
invocation.

Save/Restore Processing
In using the Save/Restore function for global variables (GLOBALV SAVEC, SAVET, GETT, and GETC), keep
in mind that a separate VSAM I/O is required for saving each variable, regardless of whether you specify
groups of variables using a wildcard character (*). The Save/Restore VSAM data set uses local shared
resources (LSR) by default. Restore operations can be buffered together, resulting in reduced I/O. Save
operations use VSAM writes, which are not deferred with LSR. If you consider using the deferred write
(DFR) performance option, see Chapter 9, “Tuning for VSAM,” on page 81 for a discussion of the
potential risk of losing data contained in the buffers in the event of an abnormal end of the NetView
program.
Do not save every global variable each time it is used in a command procedure. Save only the most critical
global variables in that manner. The processing required for GLOBALV SAVE and RESTORE is an order of
magnitude greater than the processing for GLOBALV GET and PUT.
Only the first 31 characters of a global variable name are contained in the VSAM key. Save/Restore
processes that specify wildcard names and access variables longer than 31 characters can result
in sequential searching, depending on how many such variables are duplicated within the first 31
characters.
Variables whose names and values are together over 4000 bytes are too long for DSISVRT and may be
saved to a sequential file - see samples CNMS8053 and CNMS8054 for more information. Such scenarios
take additional time to process, depending on the number of these variables. Limit the usage of these
variables with the Save/Restore function.

Chapter 4. Tuning for Command Procedures 35

36 IBM Z NetView: Tuning Guide
Chapter 5. Tuning for the Hardware Monitor

The hardware monitor is the component responsible for managing host and network problem information.
The hardware monitor manages this information using a filtering mechanism that controls data recording
and the generation of alerts to the operator.

Tuning Techniques
Following are the major tuning techniques for the hardware monitor, arranged in order of expected effect
on performance, with the most important tuning considerations listed first. These are described in detail
in this chapter.
1. Use the ESREC and AREC filters to control what data is logged to the hardware monitor database
as events, statistics, and alerts. For records that are automated using the NetView automation table,
consider blocking the recording filters from the automation table. See “Hardware Monitor Filters” on
page 37.
2. Use the ALCACHE statement to specify the number of alerts to be kept in storage. See “Using the
NPDA.ALCACHE Statement” on page 39.
3. For environments with more than 10 alerts a minute, use the viewing filter to minimize the number of
Alerts-Dynamic panels that are updated for each alert. See “Alerts-Dynamic Panel” on page 39.
4. Reorganize the VSAM database frequently to avoid control interval (CI) and control area (CA) splits.
Use the LISTCAT command to determine whether splits are occurring. You can use this information to
reorganize the database if necessary. See “ VSAM Database Maintenance” on page 88 and “LISTCAT
Command” on page 84.
5. Use the HMSTATS command to monitor the amount of hardware monitor work being done on your
system. HMSTATS displays event and alert workload counters, and statistics about the alert cache and
alerts dynamic (ALD) screen update processing. See “HMSTATS Command” on page 41.
6. For environments in which alerts are forwarded to a focal point host, use the LU 6.2 method of
forwarding alerts where possible, rather than LUC alert forwarding or the alert notification forwarding
mechanism that is based on message BNJ146I. See “Using the NPDA.ALERTFWD Statement ” on page
41.
7. Use the RATE statement to stop database logging of rapidly recurring events for a resource. See
“NPDA.RATE Statement Initialization Specifications” on page 46.
8. Code NPDA.MACRF=DFR in the CNMSTUSR or CxxSTGEN member, instead of using the LSR option. See
“Local Shared Resources (LSR) and Deferred Write (DFR)” on page 81.
Refer to “Automating Hardware Monitor Records” on page 13 and “Filtering Hardware Monitor Records”
on page 13 for more information about alert automation.

Hardware Monitor Filters

The hardware monitor uses several filters in processing statistical records and events from the network.
Use the SRFILTER command to set the following filters:
Event and statistical recording (ESREC) filter
Determines which event and statistical records are recorded in the database. Unimportant events can
be filtered using ESREC (example: NPDA SRF ESREC BLOCK xxxx).
Alert recording (AREC) filter
Determines which event records are also recorded as alerts.
Operator (OPER) filter
Determines which alerts produce messages BNJ030I and BNJ146I for the authorized receiver.

© Copyright IBM Corp. 1997, 2022 37

 Route (ROUTE) filter
Determines whether alerts are to be forwarded to the alert focal point (provided an alert focal point
exists).
Color (COLOR) filter
Determines the color in which an alert is displayed when the alert is presented on the Alerts-Dynamic,
Alerts-Static, or Alerts-History panels.
TECROUTE filter
Sets a filter for converting alerts to Event Integration Facility (EIF) events and forwarding the events
to the designated event server. An alert must pass the ESREC and AREC filters before the TECROUTE
filter is applied to the alert.

Use the SVFILTER command to determine which alerts can be viewed by a particular operator, as well as
controlling viewing of the Total Events and Total Statistics panels.
Refer to the IBM Z NetView Command Reference Volume 1 (A-N) for syntax and use of the SRFILTER and
SVFILTER commands.
Effective use of filters can have a significant effect on the hardware monitor resource requirements. You
can use many criteria to filter data. Your environment determines which are appropriate.
You can automate records using the MSUSEG and HIER conditions in the NetView automation table. If
you do not need to record automated records to the hardware monitor VSAM database, you can save
processing time by blocking the recording filters (ESREC, AREC, OPER, and ROUTE) in the automation
table. Refer to “Automating Hardware Monitor Records” on page 13 and “Filtering Hardware Monitor
Records” on page 13 for more information about automating problem records and blocking recording
filters.
Figure 9 on page 38 represents the hardware monitor filter structure.

Figure 9. Hardware Monitor Database and Filters

38 IBM Z NetView: Tuning Guide

Alerts-Dynamic Panel
For each alert that it processes, the hardware monitor checks to see if operators are on the Alerts-
Dynamic panel. For each operator on the Alerts-Dynamic panel, the hardware monitor checks the
operator's viewing filter to determine if the operator's panel should be updated. The processing required
for Alerts-Dynamic panel updates can be significant if several operators are on the Alerts-Dynamic panel,
and each must be updated with every alert. Minimize the average number of Alerts-Dynamic panels that
are updated for each alert using the viewing filter. If your installation has more than 10 alerts per minute
or a large number of Alerts-Dynamic panels active, use the viewing filter aggressively.
The following are considerations to reduce the amount of processing required for Alerts-Dynamic panel
updates:
• Before you leave your workstation, exit the Alerts-Dynamic panel. It is not a productive use of system
resources to update an unattended panel.
• If you are on the Alerts-Dynamic panel and you experience a burst of panel updates (a rapidly rolling
screen). By reentering the ALD command to go back to the Alerts-Dynamic panel, the screen will be
updated with only the most recent alerts.
The HMSTATS command displays statistics about ALD screen update processing on your system,
including the average number of screen updates per alert (UPDATES/ALERT) and the total view filtering
percentage (% FILTERED). Refer to “HMSTATS Command” on page 41 for more information.

Using the NPDA.ALCACHE Statement

You can use the NPDA.ALCACHE statement in the CNMSTYLE member to improve hardware monitor
performance, if your NetView operators do the following activities:
• Receive many alerts (either at steady rates or in bursts)
• Frequently monitor the Alerts-Dynamic panel
The NPDA.ALCACHE statement is part of the CNMSTYLE initialization member and you can specify the
number of alerts kept in storage. By keeping alert records in storage, all VSAM I/O to retrieve those alerts
from the hardware monitor database is eliminated. This allows the Alerts Dynamic Panel to be updated
without the additional overhead of the VSAM I/O. Also, the hardware monitor main task, BNJDSERV, uses
less processor time when alerts are kept in storage through ALCACHE. In general, the more storage you
allocate for alerts, the better the hardware monitor's performance.
Table 1 on page 40 shows examples of how the NPDA.ALCACHE statement can affect the performance
of your hardware monitor. Use the information to determine the appropriate NPDA.ALCACHE statement
for your system. This table assumes that one alert requires 500 bytes of storage.

Chapter 5. Tuning for the Hardware Monitor 39

Table 1. Determining the Best ALCACHE Statement for Your System
System Usage ALCACHE Storage Allocated Performance Comments
Characteristic Statement Benefit
Hardware monitor ALCACHE Storage that is Performance benefit The storage that is allocated
performance is WRAPCNT allocated depends for monitoring depends on the amount
important because on the alert wrap the Alerts-Dynamic of alert-cache storage
your NetView count. You can panel for both acceptable to your system.
operators frequently specify the alert bursts of alerts and
monitor the Alerts- wrap count with the a steady stream.
Dynamic panel. W(wrap) statement
in the CNMSTUSR
member, or with the
SWRAP command.
For example, if you
specify W AL 50,
then 25000 bytes of
storage (500 bytes
x 50 alerts) are
allocated for alerts.
NetView operators ALCACHE None None If the Alerts-Dynamic panel
rarely monitor NONE is rarely used, putting alerts
the Alerts-Dynamic into storage does not help
panel. your performance. Therefore,
do not use storage for alerts.
Your system receives ALCACHE 500 bytes x 10 Performance benefit Although you can save on
an average of 10 10 alerts = 5000 bytes during steady storage and performance
alerts per minute and of storage states. during steady states,
bursts of up to 60 performance during bursts is
alerts at once. not as good as it would be
if you defined a larger alert-
cache size.
Your system receives ALCACHE 500 bytes x 60 Performance benefit Performance is improved
an average of 10 60 alerts = 30000 during both alert during bursts, but more
alerts per minute and bytes of storage bursts and steady storage is used.
bursts of up to 60 states.
alerts at once.
System utilization ALCACHE None None The performance gained by
percentage is low. NONE keeping alerts in storage
is minimal because system
utilization is low. Therefore,
do not use storage for alerts.
Your system can ALCACHE 500 bytes x 10 Performance benefit The storage allocated
spare only a small 10 alerts = 5000 bytes during steady depends on the amount
amount of storage for of storage states. of alert-cache storage
alerts. acceptable to your system.

The HMSTATS command displays the current ALCACHE setting and statistics on the alert cache usage.
A good ALCACHE setting will have a "% SATISFIED" value near 100% without wasting storage. Refer to
“HMSTATS Command” on page 41 for more information.

40 IBM Z NetView: Tuning Guide

Using the NPDA.ALERTLOG Statement
The NPDA.ALERTLOG statement in the CNMSTYLE member enables you to make a VSAM performance
versus processor time decision in regards to reading and writing records to the hardware monitor VSAM
data base. With RSTDRANG, alert records are logged in one key range in the data base, and non-alert
records are logged in another key range in the data base. With RANDRANG, the default value, alerts are
logged throughout the data base, they are not restricted to a single key range. The ALERTLOG statement is
processed only for new (empty) data bases, and it is ignored for existing data bases. Performance is quite
comparable for RSTDRANG and RANDRANG in both small databases or until the wrap count is exceeded.
When the wrap count is exceeded or in databases with more than 10,000 alerts being held, Hardware
Monitor initialization time will take up to four times longer if RSTDRANG is coded.
Note: Hardware Monitor does not support key-ranged databases. Unpredictable results can occur.

Using the NPDA.ALERTFWD Statement

The following methods provide alert forwarding:
• Using an LU 6.2 session.
• Using an LUC session.
– Does not support intermediate node (nested) focal points.
• Converting the alert to a message (BNJ146I) and then forwarding the message over a NetView-NetView
task (NNT) to the focal point. The message is then converted back to an alert.
The NPDA.ALERTFWD statement in the CNMSTYLE member controls which alert forwarding method is
used. Forwarding over LU 6.2 sessions is extremely beneficial for hardware monitor performance for
networks that incorporate forwarding alerts across an intermediate focal point.
CPU usage is less than half when implementing LU 6.2 intermediate alert forwarding instead of converting
the message BNJ146I. This CPU reduction is recognized both in steady state and bursts of alerts. When
an intermediate focal point is not used, LU 6.2 alert forwarding provides comparable performance to LUC
forwarding.
Both the LU 6.2 and the LUC methods reduce VSAM recording at the host and have other usability
advantages. The LU 6.2 alert forwarding method is the preferred method. For additional information refer
to the IBM Z NetView Installation: Configuring Additional Components and the IBM Z NetView Automation
Guide.

HMSTATS Command
HMSTATS is an unsupported, internal serviceability tool that is helpful in tuning the Hardware Monitor. Use
the HMSTATS command to monitor the amount of hardware monitor work being done on your system.
HMSTATS displays event and alert workload counters, and statistics about the alert cache and alerts
dynamic (ALD) screen update processing. Figure 10 on page 42 shows a sample of output generated
from the HMSTATS command.

Chapter 5. Tuning for the Hardware Monitor 41

 HARDWARE MONITOR STATISTICS DISPLAY
********* RECEIVED COUNTS *********
TOTAL TRAFFIC: 289792
FROM EP: 0
RECORDED: 274545
RECORDED (GMFALERT): 0
********** FILTER COUNTS **********
TOTAL LVL1% LVL2% LVL3% LVL4% LVL5%
EVENTS/STATS: 277528 0 1 78 18 0
ALERTS: 65000 % NON-GENERIC: 21%
OPER: 0
ROUTE: 0
********** MISCELLANEOUS **********
EXTERNAL LOG: 274529 RATE VALUE: 12
CORRELATORS: 94238 PURGE IN PROGRESS: NO
*********** ALERT CACHE ***********
ALERT CACHE -
ALCACHE: WRAPCNT ALERT WRAPCNT: 112
TABLE ENTRIES: 112 TABLE STORAGE: 46K
REQUESTS: 842908 RETRIEVALS: 425701
% SATISFIED: 100%
*********** ALD COUNTS ************
NUMBER OF ALDS: 8 ALDS BEHIND: 1
TOTAL UPDATES: 11334 TOTAL FILTERED: 414421
UPDATES/ALERT: 0.2 % FILTERED: 97%
OPERID DOMAIN BEHIND UPDATES FILTERED FILTER%
------ ------ ------ ------- -------- -------
VOICE CNM12 0 67 64324 99%
KIOSK CNM12 0 365 47505 99%
BIGSGS CNM12 0 891 14894 94%
OPER20 CNM12 0 119 3038 96%
OPER02 CNM12 0 201 2857 93%
OPER12 CNM12 0 38 3076 98%
DAVE CNM12 1252 56 1544 96%
ABRAHAM CNM12 0 543 1085 66%
END OF HARDWARE MONITOR STATISTICS DISPLAY

Figure 10. Example of Output from HMSTATS Command

The HMSTATS command also has a RESET option, which sets all of the workload counters to zero.
The following information is displayed with the HMSTATS command:
RECEIVED COUNTS:
TOTAL TRAFFIC
The total number of items that have been received by the hardware monitor.
FROM EP
The total number of alerts received that were forwarded from NetView entry points.
RECORDED
The total number of items that have been recorded to the hardware monitor database.
RECORDED (GMFALERT)
The total number of items that have been recorded for GMFHS.
FILTER COUNTS:
EVENTS/STATS
The total number of events and statistics which have been recorded to the hardware monitor
database.
Level 1–5 %
The percentage of the events/statistics records that were written at the given level in the resource
hierarchy.
ALERTS
The total number of alerts which have been recorded to the hardware monitor database.
% NON GENERIC
The percentage of the alerts recorded which are not in generic format.
OPER
The total number of alerts which passed the OPER filter to generate messages BNJ030I and BNJ146I.

42 IBM Z NetView: Tuning Guide

ROUTE
The total number of alerts which passed the ROUTE filter and will be forwarded to the alert focal
point.
MISCELLANEOUS:
EXTERNAL LOG
The total number of external log records that have been written.
CORRELATORS
The total number of correlators received.
RATE VALUE
The value used for the RATE statement (in seconds) in the CNMSTYLE member.
PURGE IN PROGRESS
Indicates whether a database purge is in progress.
ALERT CACHE:
ALCACHE
The value coded on the NPDA.ALCACHE statement in the CNMSTYLE member (or the default value),
which is NONE, WRAPCNT, or a numeric value.
ALERT WRAPCNT
The current alert wrap count value.
TABLE ENTRIES
The number of entries in the alert cache table.
TABLE STORAGE
The number of kilobytes of storage allocated for the alert cache table.
REQUESTS
The number of attempts made to retrieve an alert from the alert cache.
RETRIEVALS
The number of successful attempts to retrieve an alert from the alert cache.
% SATISFIED
The percentage of alert cache requests that were satisfied. A request can be satisfied by either
retrieving the alert, or by determining that no next alert exists.
ALD COUNTS:
NUMBER OF ALDS
The number of operators that have an alerts dynamic (ALD) display active.
ALDS BEHIND
The number of ALD screens that are behind. An ALD screen is considered behind if there are alerts
recorded to the database that have not yet been checked against the operator's VIEW filter.
TOTAL UPDATES
The total number of times an ALD screen has been updated with an alert. This is a cumulative count
for all ALD screens, including those that are no longer active.
TOTAL FILTERED
The total number of times an alert was blocked by an operator's VIEW filter. This is a cumulative count
for all ALD screens, including those that are no longer active.
UPDATES/ALERT
The average number of ALD screens that are updated for each alert. This number is calculated by
dividing the total number of times an ALD screen has been updated with an alert (TOTAL UPDATES)
by the total number of alerts that have been recorded to the database (the ALERTS field in the FILTER
COUNT section). This result is intended to be a long running average, and does not account for alerts
that have been recorded to the database but not yet checked against operator VIEW filters.
% FILTERED
The average VIEW filtering percentage. This number is calculated by dividing the TOTAL FILTERED
field by the sum of the TOTAL UPDATES and TOTAL FILTERED fields.

Chapter 5. Tuning for the Hardware Monitor 43

 For each operator that has an ALD screen active:
OPERID
The operator's ID.
DOMAIN
The domain the operator is in.
BEHIND
The number of alerts which have been recorded to the database, but have not yet been checked
against the operator's VIEW filter.
UPDATES
The number of alerts that have passed the operator's VIEW filter and been sent to the operator's ALD
screen.
FILTERED
The number of alerts that have been blocked by the operator's VIEW filter.
FILTER %
The operator's VIEW filtering percentage. This number is calculated by dividing the FILTERED field by
the sum of the UPDATES and FILTERED fields.

Using the NPDA.DSRBO Statement

For the hardware monitor, the solicited data services request block (DSRBO) value specifies the projected
number of concurrent user requests for services from the BNJDSERV data services task. To determine
the number of DSRBOs that BNJDSERV needs on this host, consider the number of cross-domain
conversations in which this host can be involved at any given time, and the number of operators
performing distributed database retrieval from this host.
You can set the DSRBO value using the NPDA.DSRBO statement in the CNMSTUSR or CxxSTGEN member.
The default value is 5. You can use the DSRBS command specifying the BNJDSERV data services task
to display the current use of the hardware monitor data services request blocks (DSRBs). Refer to “Data
Services Request Blocks (DSRBs)” on page 97 for more information about tuning the DSRB allocation for
a data services task and for more detail about the DSRBS command.

Wrap Counts
Events, statistical data, and alerts are all logged to the hardware monitor database. The defaults for the
number of event and statistical records that are logged are 25 records for each resource. After 25 records,
the data is wrapped. For alerts, the number of records logged before wrapping is 100. You can decrease
the default values to make the hardware monitor database smaller. Use the SWRAP command to change
the wrap count of the records.

SWRAP Command
The SWRAP (SW) command establishes the number of event or statistical records to be retained for a
specified resource or the total number of alert records to be retained on the hardware monitor database.
You can issue the command only for resources against which data has been logged on the hardware
monitor database.
When this command requests a reduction in wrap count, the oldest records are deleted immediately. If
the wrap count is small, it might appear that the oldest record is not being wrapped off, because the new
set of records fits on the panel without deleting the old records.
For more information about the SWRAP command, see the IBM Z NetView Command Reference Volume 1
(A-N) or the NetView online help.
Note: The SWRAP command can result in loss of error data.

44 IBM Z NetView: Tuning Guide

 Example: Using the SWRAP Command
The following SWRAP command sets the event wrap count to 5 for resource UNIT1:

SW EV 5 N UNIT1

Initialization Specifications
To set wrap counts at initialization for resources that are not yet in the database, use the NPDA.W
statement in the CNMSTUSR or CxxSTGEN member. The NPDA.W statement does not alter the wrap count
for resources already existing on the hardware monitor database. The NPDA.W statement assigns initial
wrap values when the first error record for a particular resource is received.
The following WRAP command sets the wrap count to 10 for event records for a line:

W EV 010 LINE

Note: An alert wrap count that is too low can trigger the NPDA.R statement.
For more information about the NPDA.W and NPDA.R statements, see the IBM Z NetView Administration
Reference.

Error-to-Traffic (E/T) Ratio Thresholds

The hardware monitor calculates ratios of temporary errors to traffic. The hardware monitor determines
this ratio for statistical data that it receives regarding communication lines attached to a 37x5 and other
channel-attached controllers. This ratio provides a warning when a line might be experiencing many
temporary errors. The hardware monitor provides this warning by issuing an event with the description as
the following example shows:

ERROR TO TRAFFIC RATIO EXCEEDED

The hardware monitor generates an event notification whenever it receives statistical data in which the
ratio of temporary errors to traffic is greater than the threshold values.
The error-to-traffic (E/T) threshold defaults to a value of 3% for communication lines and to 1% for all
channel-attached communication controllers.
The default threshold of 3% is not appropriate for all lines. Lines vary widely in the ratio of temporary
errors to traffic. Some lines can have a normal range of 3% to 11%. For other lines, the threshold should
be at least 15% so that an event is generated for an abnormal condition only. Not only do meaningless E/T
ratio events clutter up the hardware monitor events panels, they also degrade performance. More data is
sent to the hardware monitor database when an event notification is generated from statistical data.
You can change the E/T thresholds in the hardware monitor by issuing the SRATIO operator command and
specifying the thresholds in the hardware monitor initialization program.

SRFILTER and SRATIO Commands

The SRATIO operator command changes the threshold for any resource that has existing statistical data
on the hardware monitor database. Changes to the thresholds are not maintained if the database is
reinitialized. You might want to write a command procedure containing SRATIO commands for all lines
where the default threshold is not appropriate. You can then execute this command procedure when you
reinitialize the hardware monitor database.
When you specify a line name in the command, all controllers on that line use the new threshold you
supply. The SRATIO command in the following example changes the E/T threshold for a line named
LINE1234 to 15.0%.

NPDA SRATIO 150 N LINE1234

The value 150 means 15.0%.

Chapter 5. Tuning for the Hardware Monitor 45

 Use the following SRATIO command to prevent alerts from being generated based on statistical data for
a given line. The SRATIO command in the following example prevents alerts from being generated for
LINE1234.

NPDA SRATIO OFF N LINE1234

Use the following hardware monitor SRFILTER (SRF) commands to disable thresholds for all resources in
your network, for example, to display no warning facility for statistical data.

NPDA SRF ESREC BLOCK C 044E0

NPDA SRF ESREC BLOCK C FEE40
NPDA SRF ESREC BLOCK C FEE64
NPDA SRF ESREC BLOCK C FEF45
NPDA SRF ESREC BLOCK C FFD45
NPDA SRF ESREC BLOCK C FFD46

The SRFILTER commands in this example prevent the logging of events and alerts based on exceeding E/T
ratios. The statistical data is still logged.

As with all set recording filter (SRFILTER) commands, the filter settings are lost when you shut down
the NetView program. Use a command procedure to execute your SRFILTER commands each time the
NetView program is activated, or code your initial automation table to set your filters when the hardware
monitor is initialized.
If you place SRFILTER commands in a NetView initialization command list, you need to add a short timer
delay, perhaps 1 or 2 minutes, to ensure that the hardware monitor initialization has been completed
before the command list is executed. Refer to the sample NetView automation table DSIDNMAT
(CNMS7003) for information on how to invoke filters during hardware monitor initialization.

RATIO Statement Initialization Specifications

You can also specify thresholds for the hardware monitor at initialization using the NPDA.R statement
in the CNMSTUSR or CxxSTGEN member. These thresholds do not change for resources that already
have statistical data logged on the hardware monitor database. Therefore, for resources already on the
database, the new thresholds do not take effect until the next time you reinitialize the database. If you
periodically reinitialize the database, specify E/T thresholds in the NPDA.R statement.
To change thresholds for the hardware monitor, add an NPDA.R statement in the CNMSTUSR or CxxSTGEN
member for each line.
The RATIO command in the following example changes the threshold for a line named LINE1234 to
15.0%.

R LINE LINE1234 150

The value 150 means 15.0%. The resource type (LINE) is required in a RATIO statement.

NPDA.RATE Statement Initialization Specifications

You can also specify the maximum rate at which events can be logged to the hardware monitor database.
The purpose of this function is to stop database logging of repetitive events for a resource, such as
alert flooding or alert streaming. The RATE function compares the time between an event being wrapped
off the database and the new event record being recorded on the database. A filter is set to block the
recording of events from the resource if the difference is less than the time specified on the NPDA.RATE
statement in the CNMSTYLE member. The NPDA.RATE statement filters by resource name. When the filter
is set, the BNJ045I message is issued. Determine why the filter was set and correct the cause of the
stream of alerts. The filter entry can then be deleted.
The preferred rate is one event per second for the predominant wrap count. This works out to 25 seconds
for the default wrap count of 25. If an NPDA.RATE statement is not coded in the CNMSTYLE member, the

46 IBM Z NetView: Tuning Guide

 rate value is set to zero. A zero value turns off the RATE function. See the IBM Z NetView Administration
Reference for more information about coding these statements.
Note: An event that is blocked by a recording filter set by the NPDA.RATE statement is not sent to
automation. If automation is required for these alerts, code an NPDA.AUTORATE statement in the
CNMSTUSR or CxxSTGEN member.

Event/Automation Service
Set the connection mode in the alert adapter configuration file member IHSAACFG and in the message
adapter configuration file member IHSAMCFG to connection_oriented for the event delivery method. The
default value is connection_less which specifies that a new connection should be established (and ended)
for each event that is sent. Connection_oriented specifies that a connection should be established when
the adapter is initialized and maintained for all events sent. A new connection is established only if the
connection is lost. The connection is ended when the adapter is terminated.
Make the same change on the designated event server desktop. When creating your desktop rule base
files, change tec_forward.conf in TEC_RULES to connection_oriented.
Set the trace_level in the Event/Automation Service initialization file member IHSAINIT to OFF.

Chapter 5. Tuning for the Hardware Monitor 47

48 IBM Z NetView: Tuning Guide
Chapter 6. Tuning for the Session Monitor

The session monitor component of the NetView program collects information about Systems Network
Architecture (SNA) sessions. The base data about a session is called session awareness (SAW) data. See
“SAW Data” on page 50 for more information. Other types of data that can be collected about sessions
include:
• Accounting data
• Availability data
• Session path information unit (PIU)
• Network control program (NCP) trace data
• Response time monitor (RTM) data

Major Tuning Techniques for the Session Monitor

Following are the major tuning techniques for the session monitor, arranged in order of expected effect on
performance, with the most important tuning considerations listed first. These are described in detail in
this chapter.
1. Evaluate the requirements for and use of session history data at your installation. Filter unnecessary
data using either the KCLASS DASD parameter or the DASD sense code filter. You can use sample
CNMSJM10 (described in the IBM Z NetView Installation: Configuring Additional Components) to
analyze your session monitor database and determine which sense codes occur most frequently.
Session ends that are filtered do not require VSAM I/O activity. See “DASD Option” on page 59 and
“DASD Filtering” on page 61.
2. Evaluate the requirements for and use of session awareness data at your installation. Filter
unnecessary data using either the VTAM SAW filter (in ISTMGC10) or the NetView SAW filter, which
uses the KCLASS SAW option. Use of the VTAM SAW filter is preferable, because it eliminates
NetView processing for filtered sessions. See “SAW Data” on page 50 and “SAW Option” on page
57.
3. Purge and reorganize the database off-shift on a regular basis to reclaim free space in the database.
If you can discard all of the session history data in the database, clear the session monitor database
while the NetView program is active using the RESETDB command. Clearing the database with
RESETDB is faster than purging and reorganizing the database. See “Managing the Session Monitor
Database” on page 61.
4. Use selective trace rather than global trace for PIU tracing. Trace only the sessions for which
you need trace data. See “Estimating Storage Usage” on page 115 to estimate the data's storage
requirements. See “Trace Data” on page 52 and “KEEPPIU Parameter” on page 55.
5. Use KCLASS KEEPSESS values greater than zero only for session pairs or groups of sessions with a
high incidence of session ends to keep those sessions from filling up the database. The KEEPSESS
option requires additional I/O processing for session ends. See “KEEPSESS Option” on page 60 and
“DGROUP Option” on page 61.
6. If you do not use the PURGE command or the PURGEDB command list to manage your VSAM
database, set the PURGE parameter in AAUPRMLP to DASD. Otherwise, use the default setting of
SPEED. See “PURGE Parameter” on page 62.
7. Evaluate the requirements for, and use of, availability data and accounting data at your installation.
Filter unnecessary data using the SESSTATS parameter on the INITMOD statement. See “AVAIL
Option” on page 59.
8. Use the SESSMDIS command to monitor the amount of session monitor work being done on your
system. SESSMDIS provides session counts, storage by data type, and traffic counts. See “SESSMDIS
Command” on page 63.

© Copyright IBM Corp. 1997, 2022 49

 9. Minimize processing during network activation by starting the session monitor after the network is
active (a session monitor warm start). Fewer SAW buffers are required to get session awareness
data for the network, and PIU data for control sessions is not sent. See “SAW Buffer Allocation and
Tuning” on page 50.
10. If you are collecting PIU trace data or using accounting, tune the PIU buffer allocation to achieve
buffer fullness between 80% and 95%. See “Trace Data” on page 52.
11. Set the LUCOUNT parameter to the number of logical units (LUs) in your network. The value of
LUCOUNT is used to optimize control block search algorithms and allocate session-related storage.
See “LUCOUNT Parameter” on page 66.
12. If you are collecting PIU trace data, use KEEPPIU values that are even divisors of 84 to avoid wasting
storage. See “KEEPPIU Parameter” on page 55 and “KEEPPIU Option” on page 58.
13. Use different KEEPPIU values for each keep class where possible to improve the locality of reference
for PIU data. See “KEEPPIU Option” on page 58.
14. If you issue the COLLECT RTM command to gather response time data from PUs with the RTM
feature, issue separate COLLECT commands for subsets of the PUs. Collecting from all of the PUs
at once can cause a large amount of request units to flow in the network at once. Stagger separate
COLLECT commands to spread out the flow of RUs over a longer period of time. See “RTM Data
Collection” on page 66.
15. Use the CANCEL command only as a last resort. If CANCEL must be issued, try to close the active
NLDM VSAM database (AAUVSPL or AAUVSSL) by issuing the SWITCH AAUTSKLP,T command. This
does not actually do a switch, but closes the active VSAM database. If that is not effective, issue the
NetView STOP FORCE command for each active VSAM task. If it is necessary to use the MVS CANCEL
command to bring down NetView and you have DFR specified, it might be necessary to delete and
redefine the affected database. See “Local Shared Resources (LSR) and Deferred Write (DFR)” on
page 81.

SAW Data
SAW data gives session status, session partners, and configuration data about these types of sessions:
• LU-LU
• SSCP-LU
• SSCP-PU
• SSCP-SSCP
• CP-CP

SAW data collection requires 310 to 500 bytes per session, depending on whether the session is in the
same domain, cross-domain, or cross-network.
The session monitor does not collect any other data about a session (for example, RTM or trace data) if it
does not have SAW data for the session.
You can keep SAW data selectively using either the SAW option of the KCLASS statement, the VTAM
SAW filter, or the SAW Filter Exit (DSIEX20). See “SAW Option” on page 57 for more information about
session awareness filtering.

SAW Buffer Allocation and Tuning

Define the SAW buffer allocation using the NLDM.SAWNUM and NLDM.SAWSIZE statements in the
CNMSTUSR or CxxSTGEN member.
SAW buffers are allocated in VTAM private storage. VTAM allocates the buffers above the 16 MB line.
The DSIAMLUT task receives the SAW buffers from VTAM and forwards them to the AAUTSKLP task for
processing.

50 IBM Z NetView: Tuning Guide

 The SAW buffering ratio, the average number of SAW notifications sent per buffer, helps to characterize
VTAM SAW buffering behavior. You can calculate the SAW buffering ratio for your system with the
following formula, using fields that are displayed with the SESSMDIS command. To calculate the SAW
buffering ratio over an interval using two observations of the SESSMDIS command, take the differences in
the session start, session end, and SAW buffer counts and use the following formula to calculate.

(session starts + session ends)

SAW buffering ratio = ----------------------------------------------
SAW buffers

See “SESSMDIS Command” on page 63 for more information.

An understanding of how VTAM buffers session awareness notifications is helpful in tuning the SAW buffer
allocation.

Tuning the SAW Buffer Allocation

VTAM SAW buffering behavior occurs in the following ways:
• During steady state periods when the session monitor is keeping up with the SAW data sent from VTAM,
the SAW buffering ratio is close to 1.
• During a session monitor warm start, VTAM packs SAW buffers.

A data space (ISTNMSDS) is used for transferring SAW data between VTAM and the session monitor.
The following VTAM start options control VTAM SAW buffering behavior. Refer to the appropriate VTAM
publication for more information on the VTAM start options.

SAW Data Space Packing Factor

The SAW data space packing factor (start option SAWMXQPK or ISTRACON constant RACSAWPK)
calculates the number of SAW data space buffers to queue before packing SAW buffers. The packing
threshold equals SAWMXQPK times the number of SAW buffers in VTAM private storage (defined by the
SAW BUFNUM parameter in AAUPRMLP). The default value is zero. The default value causes VTAM to
pack SAW buffers if the data space is not empty. This occurs if an earlier SAW buffer sent by VTAM
was not received by the session monitor. This packing behavior improves performance because as traffic
increases, the buffering efficiency increases. Use the default value of zero for SAWMXQPK.

SAW Data Space Limit Factor

The SAW data space limit factor (start option SAWMAXDS or ISTRACON constant RACSAWLM) calculates
the maximum buffer limit for the SAW data space. The maximum number of SAW data space buffers
equals SAWMAXDS times the number of SAW buffers in VTAM private storage (defined by the SAW
BUFNUM parameter in AAUPRMLP). The default value for SAWMAXDS is 100. If the session monitor
is unable to keep up with the SAW traffic, VTAM queues SAW buffers in the SAW data space. VTAM
continues adding SAW buffers to the data space until the maximum buffer limit is reached.

SAW Buffer Limit

The SAW buffer limit (start option MXSAWBUF or ISTRACON constant RACMXBUF) sets the maximum
number of SAW buffers that can be allocated in VTAM private storage. SAW buffers back up in VTAM
private storage only after the SAW data space has reached its maximum buffer limit (controlled by
SAWMAXDS). When this happens, VTAM tries to conserve SAW buffers by placing only SAW data
associated with termination into the buffers. When the SAW buffer limit is reached, VTAM ends the
session awareness function, releasing the SAW data space and the allocated buffers. When buffers back
up in VTAM private storage, extra VTAM processing time is needed to conserve SAW buffers and allocate
new buffers from VTAM private storage.

Chapter 6. Tuning for the Session Monitor 51

 Considerations for Defining the SAW Buffer Allocation
In defining the SAW buffer allocation, data space, and limits, you can control the maximum amount of
SAW data that is allowed to be kept in the system prior to the session monitor processing the data. Use
the data space to keep unprocessed SAW data and avoid letting SAW data back up into VTAM private
storage. The following are considerations.
• Use the default value of zero for SAWMXQPK. This gives the best SAW buffering ratio.
• Use a BUFSIZE that is a multiple of 4 K (4 K—32 K). When VTAM is packing buffers, a larger BUFSIZE
allows more SAW notifications to be sent for each buffer. You can monitor the SAW buffering ratio using
fields from the SESSMDIS command.
• Use a small value such as 2 for BUFNUM. Because the data space is used to store unprocessed buffers,
you do not need to define a large number of SAW buffers.
• Set SAWMAXDS in conjunction with the SAW BUFSIZE and BUFNUM parameters in AAUPRMLP to reflect
the amount of data space storage that you are willing to devote to a backlog of SAW data. To determine
the peak size of the SAW data space, multiply SAWMAXDS by BUFSIZE and BUFNUM. For example,
if AAUPRMLP defines two 4K SAW buffers and SAWMAXDS is equal to 100, the maximum size of the
SAW data space is 800K (2 * 4K * 100). Consider using a large number for SAWMAXDS to keep all
unprocessed SAW buffers in the data space and avoid a backup into VTAM private storage. Small values
for SAWMAXDS and MXSAWBUF result in VTAM ending the session awareness function prematurely.
• Set MXSAWBUF to a relatively low number (200 or less). If you define SAWMAXDS properly, a backup
into VTAM private storage should not occur. If a backup does occur, a small value for MXSAWBUF causes
VTAM to end the session awareness function before the backup is significant.

Trace Data
The session monitor provides two trace modes:
• Global
• Specific

Global Tracing
The global trace mode traces all sessions. The TRACESC and TRACELU parameters on the INITMOD
statement in AAUPRMLP are used to specify whether global tracing is performed. These parameters are
specified with the statements as shown in the following example.

NLDM.TRACESC=YES|NO
NLDM.TRACELU=YES|NO

Note: The default values are underlined.

TRACESC=YES indicates global tracing of SSCP/CP sessions (SSCP-LU, SSCP-PU, SSCP-SSCP, and CP-CP).
TRACELU=YES indicates global tracing of LU-LU sessions. Global tracing can require a large amount
of virtual and DASD storage. See “Estimating Storage Usage” on page 115 to determine the storage
requirements before using global trace.
You can use the NLDM TRACE DISP command to determine which sessions are being traced. Global NLDM
TRACE defaults can be changed dynamically.

Selective Tracing
If you do not want to use global trace, you can specify selective trace with the session monitor TRACE
command. For example, if you want to trace only SSCP-SSCP sessions, do the following:

52 IBM Z NetView: Tuning Guide

1. Specify TRACESC=NO and TRACELU=NO in initialization member AAUPRMLP. You can code this as
shown in the following example.

NLDM.TRACESC=NO
NLDM.TRACELU=NO

2. Issue a TRACE START command for each SSCP that will be in session with VTAM. Assuming three
other SSCPs whose names are CDRM1, CDRM2, and CDRM3, the session monitor TRACE commands
are shown in the following example.

NLDM TRACE START CDRM1

NLDM TRACE START CDRM2
NLDM TRACE START CDRM3

You can code these commands in a NetView command list. You can also code a NetView command list
that dynamically determines the SSCPs and starts the TRACE command for them. You can issue the
VTAM display command, as shown in the following example, in a command list to identify the SSCPs.

(D NET,ID=VTAM,E)

Then issue NLDM TRACE START commands in the same command list for those SSCPs.

If your network experiences a high level of session failures, consider tracing the SSCP sessions. Using the
KEEPPIU option of the KCLASS statement, keep more PIUs for the following:
• SSCP-SSCP sessions
• SSCP-PU sessions for gateway NCPs
• SSCP-LU sessions for VTAM applications

For all other SSCP sessions, keep fewer PIUs. For networks with hung terminals and protocol problems,
consider tracing LU-LU sessions.
The type of information available from PIU tracing varies depending on the session type. The following are
some examples:
• Session initiations (INITs) for SSCP-LU sessions for applications
• Cross-network session resource allocation (RNAAs, SETCVs) for SSCP-PU sessions for gateway NCPs
• Cross-domain and cross-network session initiations (CDINITs) for SSCP-SSCP sessions
Activate trace for other session types or resources when needed to reproduce a problem or to monitor
selected sessions.
The accounting function summarizes the PIU data by session. When you use the accounting function,
specified with SESSTATS=YES in AAUPRMLP, the session monitor receives and processes the same
amount of PIU trace data as if global trace had been specified. If you specify a trace for the session
with the TRACESC or TRACELU parameters or the TRACE command, the PIU data is available for viewing
and recording. Tracing does not have to be active to use the accounting function.
If you use the availability function, specified with SESSTATS=AVAIL in initialization member AAUPRMLP,
rather than the accounting function, the session monitor does not require tracing to be active. Therefore,
the processing and storage associated with global tracing can be eliminated by specifying TRACESC=NO
and TRACELU=NO in member AAUPRMLP.
The amount of host processing required for the accounting function is similar to that required for global
trace. Global trace can require considerably more storage than accounting, however, depending on the

Chapter 6. Tuning for the Session Monitor 53

 number of PIUs kept in storage for each session. See “KEEPPIU Parameter” on page 55 for information
about how to specify the number of PIUs to keep for a traced session.
Note: When you filter SAW data with VTAM, VTAM does not send PIU data for filtered sessions to the
session monitor. If you filter SAW data with the session monitor KCLASS statement instead of the VTAM
SAW filter, VTAM sends PIU data for filtered sessions to the session monitor and the data is discarded.
See “SAW Option” on page 57 for more information on session awareness filtering.
If you want to use global trace or accounting and restrict the number of sessions for which PIU data is
processed, you can use the VTAM SAW filter to filter unneeded sessions. Keep in mind that if you filter a
session, no other data can be collected for that session.

PIU Buffer Allocation and Tuning

Define trace buffers using the NLDM.PIUTNUM and NLDM.PIUTSIZE statements in the CNMSTUSR or
CxxSTGEN member.
The DSIAMLUT task receives the PIU buffers from VTAM and forwards them to the AAUTSKLP task for
processing. The storage for PIU buffers is common storage page fixed. If VTAM does not have enough
trace buffers and starts overlaying them, the NetView program issues message AAU024I indicating the
number of lost buffers. Losing PIU buffers is usually not a problem unless you are using the session
monitor accounting function or are not getting the desired trace data.
Three goals for tuning the PIU buffer allocation are:
• Maximize the number of PIUs that are sent per buffer.
• Avoid losing PIU buffers.
• Avoid over allocating buffers, which wastes storage.

VTAM sends full PIU buffers unless the NetView program solicits a buffer, as it might when processing
a session-end notification or when processing an operator request. Therefore, the optimum PIU buffer
allocation depends on the rate at which session-end notifications are processed and the frequency of
operator requests for PIU trace data.
In tuning the PIU buffer allocation, determine the optimum size of the buffers for your environment. Then
adjust the number of buffers to allocate so that PIU buffers are not lost during periods of peak activity.
The SESSMDIS command and the external log record created by the NLDM RECORD STRGDATA command
provide information that you can use to tune the PIU buffer allocation for your system. The SESSMDIS
command displays session monitor traffic data and event counters, while the NLDM RECORD STRGDATA
command writes similar information to the NetView external log. The NLDM RECORD STRGDATA
command creates SMF record type 39, subvector X'0008'. See the IBM Z NetView Command Reference
Volume 1 (A-N) or the NetView online help for information about the syntax and use of those commands.
See the IBM Z NetView Application Programmer's Guide for the format of the external log record.

Estimating PIU Buffer Fullness

Both the SESSMDIS and NLDM RECORD STRGDATA commands provide counters for the number of PIU
buffers processed and the number of PIUs processed by the NetView program. Use the counters to
estimate the fullness of the PIU buffers sent from VTAM to the NetView program using the following
steps:
1. Collect the PIU buffer counts and PIU counts (with the SESSMDIS command) from the NetView
program over a period of 1 to 2 hours during peak activity for your system.
2. Using your current PIU buffer size and the PIU buffer and PIU counts from your period of peak activity,
estimate the average percent fullness of your PIU buffers using the following formula:

(PIU count x 48)

% full = -------------------------------------------- x 100
(PIU buffer count x current PIU buffer size)

54 IBM Z NetView: Tuning Guide

 The constant 48 represents the size in bytes of the vast majority of PIUs that the NetView program
processes. Because the PIU buffer contains some buffer header information, the buffer fullness
percentage will approach but never equal 100%.
In general tune your buffer size to achieve buffer fullness of 80% or more. If your PIU buffers are less
than 80% full and storage is a constraint on your system, you can decrease the buffer size.
If your PIU buffers are 95% full or more, you might benefit from increasing the PIU buffer size. A larger
buffer size might allow you to get more PIUs sent in each buffer.
The samples contain a starting allocation of two 4 K PIU buffers. If you are in a large environment, you
might want to use four 8 K PIU buffers for your starting allocation.
In the process of tuning your PIU buffer size, adjust the number of buffers so that the total size of your
PIU buffer allocation (number of buffers multiplied by buffer size) does not vary significantly. When you
have arrived at a satisfactory PIU buffer size, you can adjust the number of PIU buffers so that you do not
lose buffers during periods of peak activity, such as during a major node recovery.
If you still receive message AAU024I (indicating lost buffers) after tuning the PIU buffer allocation, the
session monitor might not be dispatched frequently enough to sustain its workload. This is an indication
of a performance problem, where higher priority work on your system (either in the NetView Program or
in other address spaces) has high CPU usage. Raising the dispatch priority for AAUTSKLP (specified in
CNMSTASK with the PRI parameter) from the default of 8 to 5 might help relieve this problem. However,
you should try to address the underlying problem of high CPU usage with the TASKUTIL command and
system performance monitoring tools, such as RMF.

PIU Data Space

A data space (ISTNMPDS) is used to transfer PIU trace data between VTAM and the session monitor.
Using the data space does not affect selecting the proper PIU buffer size for your environment, but does
affect the number of PIU buffers that you should specify.
The VTAM PIUMAXDS start option (or ISTRACON constant RACPIULM) is the PIU data space limit factor
used to calculate the maximum buffer limit for the PIU data space. The maximum number of PIU data
space buffers equals PIUMAXDS times the number of PIU buffers in VTAM private storage (defined by the
PIU BUFNUM parameter in AAUPRMLP). The default value for PIUMAXDS is 200. If the session monitor is
unable to keep up with the PIU traffic, VTAM queues PIU buffers in the PIU data space. VTAM continues to
add PIU buffers to the data space until the maximum buffer limit is reached. VTAM then discards PIU data
until buffers become available. When the session monitor detects that PIU buffers were lost, message
AAU024I is issued.
Set PIUMAXDS in conjunction with the PIU BUFSIZE and BUFNUM parameters in AAUPRMLP to reflect the
amount of data space storage that you are willing to devote to a backlog of PIU data. To estimate the peak
size of the PIU data space, multiply PIUMAXDS by BUFSIZE and BUFNUM. For example, if AAUPRMLP
defines four 8K PIU buffers and PIUMAXDS is equal to 200, the maximum size of the PIU data space is
6400 K (4 × 8 K × 200).

KEEPPIU Parameter
The KEEPPIU parameter in AAUPRMLP specifies the number of PIUs to keep for a traced session and
affects virtual storage, processor storage, and DASD storage. The value provided in the samples is 7.
KEEPPIU affects the amount of virtual storage the session monitor uses for the period a session is active,
and the amount of DASD storage used when a session is deactivated. For each PIU kept, 50 bytes of
virtual storage are used per session.
The session PIUs are kept in a wraparound area in virtual storage while the session is active and are
stored in the database when the session ends. When the first PIU for a session is received, the entire
KEEPPIU storage for that session is allocated. Specifying a large KEEPPIU value for sessions that do not
have much traffic, such as SSCP-LU and SSCP-PU sessions, could result in wasted storage. See “KEEPPIU
Option” on page 58 for information on tailoring the KEEPPIU value for different keep classes.

Chapter 6. Tuning for the Session Monitor 55

 Use the NetView SESSMDIS command to display the amount of storage allocated for the PIU trace data.
This information is helpful in determining a good value for KEEPPIU. See “SESSMDIS Command” on page
63 for a description of the syntax and use of this command.

TRACEGW Parameter
The TRACEGW parameter in AAUPRMLP specifies whether NCP gateway trace data is to be collected for
cross-network sessions. If TRACEGW is set to YES, when the session monitor receives a session-start
notification for an NCP, it requests the NCP to collect gateway trace data for cross-network sessions that
pass through it. When cross-network sessions end, the NCP sends the data to the session monitor for
recording to the VSAM database. If you do not use this trace data for problem determination or other
purposes, you can set TRACEGW to NO and eliminate the extra processing and VSAM storage.

Keep Classes
You can use keep classes to keep selective SAW data for active sessions in addition to recording selective
session data on sessions that have ended. Keep classes can also control the amount of trace data
collected. As mentioned in “SAW Data” on page 50, if you plan to keep SAW data selectively, consider
filtering SAW data with the VTAM SAW filter.
KCLASS and MAPSESS statements are used to define the selectivity of keep classes, as shown in Figure
11 on page 57. These statements are in a separate DSIPARM data set member, which is defined by the
NLDM.KEEPMEM statement in the CNMSTYLE member.
KCLASS defines selective processing options (AVAIL, SAW, DASD, KEEPSESS, DGROUP, and KEEPPIU).
MAPSESS defines which sessions use a specified KCLASS.
KCLASS and MAPSESS definitions are loaded during session monitor initialization. You can change these
definitions dynamically using the session monitor RELOAD command to load new members. Active
sessions are not affected by the RELOAD command.
Keep-class processing is done on a session basis and options are applied when the session monitor
receives a session start. The session partner names are used as criteria to search the MAPSESS
statements until a match is found. You can use the question mark (?) and asterisk (*) for matching names
when you follow naming conventions. These are called pattern-matching or wildcard characters. The ?
wildcard character holds a place in the column, specifying that any character can be there. The * wildcard
character specifies a match with any character in the column along with any characters to the end of the
name.

Network Resource Naming Conventions

Network resource naming conventions enable keep-class coding. The following are example naming
conventions:

CDRMnnnn
CDRM names
NCPGWnnn
Gateway NCP names
NCPnnnnn
NCP names
nnnPUnnn
PU names
nnnLUnnn
SNA LU names
nnnLUnnB
Bisynchronous LU names

56 IBM Z NetView: Tuning Guide

 CICSnnnn
CICS®* applications
TSOnnnnn
TSO applications
Figure 11 on page 57 uses this naming convention to specify processing options for different classes of
sessions. The PLU name for SSCP sessions is VTAM in this figure.

SSCPSSCP KCLASS SAW=YES,DASD=YES,KEEPPIU=84

SSCPGWN KCLASS SAW=YES,DASD=YES,KEEPPIU=84
SSCPNCP KCLASS SAW=YES,DASD=YES,KEEPPIU=42
SSCPPU KCLASS SAW=YES,DASD=NO,KEEPPIU=6
SSCPCICS KCLASS SAW=YES,DASD=YES,KEEPPIU=21
SSCPOTHR KCLASS SAW=YES,DASD=NO,KEEPPIU=4
BSYLULU KCLASS SAW=YES,DASD=(DATA,FAILURES),KEEPPIU=14
SNALULU KCLASS SAW=YES,DASD=(DATA,FAILURES),KEEPPIU=12
TSOLULU KCLASS SAW=YES,DASD=NO,KEEPPIU=12
ALLOTHER KCLASS SAW=YES,DASD=(FAILURES),KEEPPIU=7
M1P MAPSESS KCLASS=SSCPSSCP,PRI=*,SEC=CDRM*
MIS MAPSESS KCLASS=SSCPSSCP,PRI=CDRM*,SEC=*
M2 MAPSESS KCLASS=SSCPGWN,PRI=VTAM,SEC=NCPGW*
M3 MAPSESS KCLASS=SSCPNCP,PRI=VTAM,SEC=NCP*
M4 MAPSESS KCLASS=SSCPPU,PRI=VTAM,SEC=???PU*
M5 MAPSESS KCLASS=SSCPCICS,PRI=VTAM,SEC=CICS*
M6 MAPSESS KCLASS=SSCPOTHR,PRI=VTAM,SEC=*
M7 MAPSESS KCLASS=TSOLULU,PRI=TSO*,SEC=*
M8 MAPSESS KCLASS=BSYLULU,PRI=*,SEC=???LU??B
M9 MAPSESS KCLASS=SNALULU,PRI=*,SEC=???LU*
M10 MAPSESS KCLASS=ALLOTHER,PRI=*,SEC=*

Figure 11. Example of KCLASS and MAPSESS Statements

The search order against the MAPSESS statements is from top to bottom. The order of the MAPSESS
statements and the accuracy of the PLU and SLU names are important. The first match is used for the
keep-class member. When a session's PLU and SLU names match a MAPSESS statement, the search ends
and the keep-class processing is determined by the KCLASS to which the MAPSESS points.
Use the session monitor DISKEEP PIU command to verify that the intended sessions are being mapped
to the desired KCLASS statements. Refer to the IBM Z NetView Installation: Configuring Additional
Components for more information about keep-class processing.

SAW Option
You can keep SAW data selectively by using either the SAW option of the KCLASS statement or the VTAM
SAW filter. If you plan to keep SAW data selectively, filter SAW data with the VTAM SAW filter. When
SAW data is filtered by the session monitor, VTAM transports the data to the session monitor, which
discards the data. Filtering SAW data with VTAM avoids the overhead of transporting the filtered data.
When sessions are filtered by the VTAM SAW filter, PIU data for the filtered sessions is not sent to the
session monitor. VTAM session awareness filtering is described in VTAM Network Implementation Guide.
If you decide to filter SAW data, use care in deciding which sessions to filter. If the SAW data for a session
is filtered, no other data can be collected for that session. Consider the following when filtering SAW data:
• SAW filtering with the session monitor KCLASS statement applies only to SSCP-LU and LU-LU sessions.
You cannot filter SSCP-PU and SSCP-SSCP SAW data for the session monitor, because session
awareness data for these sessions is required in many session monitor functions, such as the response
time monitor (RTM), gateway trace, and boundary function trace.
• The VTAM SAW filter enables you to filter SSCP-PU and SSCP-SSCP sessions. Do not filter SSCP-PU and
SSCP-SSCP session awareness data unless you are certain that you do not require this data for your
environment. If you collect RTM data using the COLLECT RTM * command, filtering SSCP-PU session
awareness for a subarea PU causes data from RTM devices in that subarea to be omitted from the
collected data.
• RTM data requires session awareness data (SAW=YES) for the related SSCP-PU, SSCP-LU, and LU-LU
sessions.

Chapter 6. Tuning for the Session Monitor 57

 • Non-RTM SSCP-LU sessions and low-priority, stable LU-LU sessions are good candidates for SAW
filtering.

DSIEX20-SAW Filter Exit

Installation exit DSIEX20 allows more granular processing of NetView session awareness (SAW) data
records. This exit allows the user to accept or discard SAW records based on their content. The exit
causes SAW data received from VTAM to be filtered by NetView Session Monitor prior to filtering
associated with a KEEPMEM defined with an INITMOD statement in AAUPRMLP containing KCLASS and
MAPSESS statements.
The sample DSIEX20 that is included with the NetView product filters out all SSCP-LU session data at
session start. The sample DSIEX20 is meant to be used as an example. It should be customized to meet
your individual needs. For more information on Exit 20, see IBM Z NetView Programming: Assembler.
Using DSIEX20 can greatly reduce the CPU and storage resources used by the Session Monitor, primarily
with the main NLDM task, AAUTSKLP. However, careful consideration should be given when coding and
using this exit. SAW data filtered at initialization might cause certain other functions within session
monitor not to have the required data needed for the function. For example, the DSIEX20 sample that is
included with the NetView product filters all SSCP-LU data. Having this data filtered at initialization would
cause RTM data to be unavailable for the filtered sessions, because RTM must have SSCP-LU data.

KEEPPIU Option
If you are tracing, tailor the number of PIUs kept for different session types to optimize processing and
control the amount of trace data kept. The KEEPPIU parameter value is determined as follows:
• If the KEEPMEM member has been coded and the session is mapped to a KCLASS, use the KEEPPIU
value coded on the KCLASS statements.
• If the KEEPMEM member has not been coded, use the KEEPPIU value coded on the INITMOD statement
in AAUPRMLP (or equivalent). The value in the samples is 7.

Where possible, each keep class should have a different KEEPPIU value. The PIU wrap areas for sessions
with the same KEEPPIU value are grouped on the same pages of virtual storage. Separating keep classes
by using different KEEPPIU values can improve the locality of reference for PIU data.
To minimize wasted storage, the values you specify for KEEPPIU should be even divisors of 84 (2, 3, 4,
6, 7, 12, 14, 21, 28, or 42). Avoid using values greater than 84. Larger values increase virtual storage
requirement of the session monitor and cause the PIU wrap area to span more than one page of virtual
storage, requiring a GETMAIN rather than using pooled storage. Small KEEPPIU values keep the virtual
storage requirement of the session monitor to a minimum.
The following are some suggested KEEPPIU values for different types of sessions:
LU-LU (APPL-APPL) sessions
42
LU-LU sessions for bisynchronous terminals
14
LU-LU sessions for SNA terminals
12 or 7 (depending on available storage)
SSCP-LU sessions for applications
21
SSCP-LU sessions for terminals
4
SSCP-PU sessions for gateway (GW) NCPs
84
SSCP-PU sessions for other NCPs
42

58 IBM Z NetView: Tuning Guide

 SSCP-PU sessions for other PUs
6
SSCP-SSCP sessions
84
CP-CP sessions
84

AVAIL Option
The SESSTATS and LOG parameters in initialization member AAUPRMLP specify whether session start and
session end records are written to the external log. The records are used to track session accounting
(availability and PIU counts) or session availability. If you do not need session accounting, which includes
PIU trace counts, specify SESSTATS=AVAIL on the INITMOD statement in member AAUPRMLP. If you
specify the availability option at initialization, you can define whether availability data should be kept
for a class of sessions using the AVAIL parameter on the KCLASS statement. You can use the RELOAD
command to dynamically change availability by keep class.
When global tracing is not requested (TRACESC=NO and TRACELU=NO) and the availability option is used
rather than accounting, the session monitor CPU utilization and working-set size are reduced.

DASD Option
Data for a session is written to the VSAM database when the session ends. This data includes SAW data
plus any optional trace and RTM data. For large networks or networks with a high incidence of session
ends, the processing required for data recording can be substantial. An important tuning consideration is
to evaluate your requirements and use of session history data. If there is history data you know you do not
need, filter it out using the KCLASS DASD option.
The following is the list of values for the DASD option of keep-class processing:
DASD=BINDFAIL
Record only if session fails during BIND (BINDF).
DASD=DATA
Record only if trace or RTM data is present.
DASD=FAILURES
Record if session fails or abnormal UNBIND occurs.
DASD=INITFAIL
Record only if session fails before BIND (INITF).
DASD=NO
Do not record any session data.
DASD=RTMDATA
Record only if RTM data is present for this session.
DASD=SESSFAIL
Record only if abnormal UNBIND occurs.
DASD=SESSNORM
Record only if normal UNBIND occurs for this session.
DASD=TRACDATA
Record only if trace data is present for this session.
DASD=YES
Always record the session data (default).
The following is an example of coding multiple options on a KCLASS DASD statement.

DASD=(BINDFAIL,RTMDATA,INITFAIL)

Chapter 6. Tuning for the Session Monitor 59

 You can code multiple conditions in a single specification. This example of the DASD option specifies that
session data is recorded:
• If the session fails during BIND (BINDF)
• If RTM data is present for this session
• If the session fails before BIND (INITF)

Filtering with the KCLASS DASD Option

By filtering data recording with the KCLASS DASD option, you can reduce host processor utilization, VSAM
I/O, and DASD storage used by the session monitor. Analyze the network problems your installation
encounters and determine the types and amounts of session monitor information needed to diagnose
the problems. For LU-LU sessions, consider using DASD=DATA, DASD=FAILURES, or DASD=RTMDATA. For
SSCP sessions for which you do not need information, use DASD=NO. For more information, see “DASD
Filtering” on page 61.
If you issue the FORCE command, the session data is recorded regardless of how you coded the DASD
parameter.

Special Relationships between the SAW and DASD Parameters

• Unless filtered by VTAM, SAW data is always kept for SSCP-SSCP and SSCP-PU sessions, no matter what
is coded on the KCLASS statement.
• If you code SAW=NO on the KCLASS statement for LU-LU sessions, only BINDF and INITF are kept as if
you coded DASD=(INITFAIL,BINDFAIL).
• If you code SAW=NO with the INITMOD statement in AAUPRMLP, SAW data is not kept for any session
regardless of the session type.

KEEPSESS Option
The KEEPSESS value for a session, specified with either the KEEPSESS parameter in AAUPRMLP or the
KEEPSESS option on the KCLASS statement, controls the number of session incidences that are recorded
on the session monitor database for a given name pair or DASD group. See “DGROUP Option” on page 61
for more information on DASD groups. The KEEPSESS value for a session is determined as follows:
• The KEEPSESS parameter on the INITMOD statement in AAUPRMLP indicates whether DASD session
wrapping is used. If you do not specify a value, the default is zero and session wrapping is not used
regardless of any KCLASS KEEPSESS values. Also, sessions will not be recorded into DGROUPs as
defined on a KCLASS statement. If you do specify a value, the value is used as the global DASD session
wrap count for sessions not mapped by MAPSESS⁄KCLASS statements and for mapped sessions having
no KEEPSESS coded.
• If the global DASD session wrap count (the KEEPSESS parameter in AAUPRMLP) is greater than zero,
the KEEPMEM member was coded, and the session is mapped into a KCLASS through a MAPSESS
statement, then the value of the KEEPSESS option that is coded on the KCLASS statement is used. If
the KEEPSESS option is not coded on the KCLASS statement, the global KEEPSESS value in the range of
1–999 is used for sessions mapping into this KCLASS.
• If the global DASD session wrap count is greater than zero, the KEEPMEM member was coded, and
the session is not mapped to a KCLASS, then the global DASD session wrap count that is coded in
AAUPRMLP is used. If the last MAPSESS statement is coded with PRI=* and SEC=* (for example,
MAPSESS KCLASS=ALLOTHER,PRI=*,SEC=*), then all sessions that do not match a previous MAPSESS
statement are mapped into the KCLASS specified (in this example, ALLOTHER).
See “Managing Database Size” on page 62 for considerations on controlling the size of the session
monitor database.

60 IBM Z NetView: Tuning Guide

DGROUP Option
The DGROUP option can be used in conjunction with the KEEPSESS option on the KCLASS statement
to control the number of session incidences that are recorded on the session monitor database for
a group of sessions. Using the KEEPSESS option alone is not effective for controlling the number of
session incidences on the database for applications, such as TSO and the NetView program, where
the primary LU (PLU) session partner name contains a counter (for example, TSO00001, TSO00002,
NETV001, NETV002). The DGROUP option can be used to group sessions together, so that the KEEPSESS
value can be applied to control the number of session incidences for the group instead of for the individual
name pairs.
DGROUP specifies the grouping characteristics of all MAPSESS sessions mapping to a KCLASS statement.
You can group sessions under a user-supplied name, or defer the DGROUP name until the session ends by
using the *PRI or *SEC values. Deferring the DGROUP name enables the definition of multiple DGROUPs
with a single KCLASS statement, with the DGROUP name being either the primary or secondary session
partner name. All defined DASD GROUPs (DGRPs) can be displayed using the NLDM LIST DGRP command.
Figure 12 on page 61 shows an example of how to use the KEEPSESS and DGROUP options to control
the number of session incidences on the database for a group of sessions. In this figure, the NetView
domain identifier is N2412, and the TSO VTAM application name is TSO12.

⋮
TSO KCLASS SAW=YES,+
DASD=FAILURES,+
KEEPSESS=200,+
DGROUP=(TSO,RENAME,PRI)
NETVIEW KCLASS SAW=YES,+
DASD=FAILURES,+
KEEPSESS=100,+
DGROUP=(NETVIEW,RENAME,PRI)
⋮
TSO MAPSESS KCLASS=TSO,PRI=TSO12*,SEC=*
NETVIEW MAPSESS KCLASS=NETVIEW,PRI=N2412*,SEC=*
⋮

Figure 12. Example of KCLASS and MAPSESS Statements Using the KEEPSESS and DGROUP Options

Managing the Session Monitor Database

The following sections describe techniques for managing the session monitor database.

DASD Filtering
You can filter session history recording using the DASD option of the KCLASS statement, which enables
you to specify recording conditions for sessions mapped to a keep class. For example, you can record only
sessions that have RTM data by specifying DASD=RTMDATA for the keep class. See “DASD Option” on
page 59 for information about the DASD option.
You can also filter session history recording based on the sense codes and reason codes that
accompany the session awareness notification. The sense code filter consists of 25 entries in DSICTMOD
(CNMS0055). Each entry consists of an 8 byte field specifying the sense and reason codes, and a numeric
field indicating the number of bytes used for comparison in the filter. By modifying DSICTMOD, you can
filter DASD recording based on up to 25 sense codes.
To analyze your session monitor database use sample CNMSJM10, these samples print a report
containing the distribution of the sense codes on the database. Use this report to determine which sense
codes occur most frequently and where to concentrate your filtering efforts.
Filter sense codes that occur frequently and are not useful in network problem determination. For
example, X'087D0001' is a normal response to an unsuccessful cross-network session setup when you
have multiple gateways and VTAM checks an incorrect gateway name first.
Refer to the IBM Z NetView Installation: Configuring Additional Components for information about deciding
which sense codes to filter, how to add sense codes for filtering, and how to end sense code filtering.

Chapter 6. Tuning for the Session Monitor 61

Managing Database Size
As sessions end, the session monitor records history data to the VSAM database. The database eventually
fills up. There are three principal methods to manage the size of the database.
1. Clear the entire database using the RESETDB command or the CLEAR parameter of the DBAUTO
command. This method is quick, but all of the session history data is lost at once, rather than in stages.
To use this method, ensure that the REUSE operand is specified for both AAUVSPL and AAUVSSL data
sets on the cluster definition in member CNMSI101.
2. Purge data for sessions that ended before a specified time by using either the PURGE command or the
PURGE parameter of the DBAUTO command. An advantage of this method is that the purge activity
can be scheduled for an off-shift period, deferring the I/O needed to delete data and minimizing
prime-shift session monitor I/O. For large databases, however, the purge activity can take a long time.
3. Use the KEEPSESS parameter or the KEEPSESS option of the KCLASS statement to control the number
of session incidences on the database for a given name pair or group of sessions (specified by the
DGROUP option). For example, if the KEEPSESS value for a name pair is 10, up to 10 incidences are
maintained on the database. When the 11th incidence is recorded, the first incidence is erased. Using
KEEPSESS requires additional I/O to delete data when sessions end. Session ends that delete data
(KEEPSESS value exceeded) require approximately three times the I/O required to record sessions that
do not delete data. See “KEEPSESS Option” on page 60 for more information about KEEPSESS.
Method 1 (clearing the entire database) performs better than the other two alternatives. However, if
method 1 does not meet your needs, method 2 (purging the database) is preferred over method 3 (using
KEEPSESS). Purging data enables you to defer session incidence deletions to off-peak hours.
Methods 2 and 3 require that you reorganize the database periodically to reclaim previously used free
space. Without periodic reorganization, the VSAM data set eventually runs out of free space, causing an
end-of-file condition. See “ VSAM Database Maintenance” on page 88 for information about reorganizing
a VSAM database.
You can control your database with KEEPSESS and purge activity in combination. If certain name pairs or
groups of sessions have a high incidence of session ends and you want to keep them from filling up your
database, use KEEPSESS to limit the session incidences for those session pairs, and purge the data for
other sessions. Because additional processing is required to control the wrap count, purge activity is more
efficient if you do not purge sessions controlled by KEEPSESS.

PURGE Parameter
The PURGE parameter in AAUPRMLP is used to control writing an end-time record to the database for use
in purge processing. PURGE=SPEED (the default) writes an end-time record. Specifying PURGE=DASD
instructs the session monitor to optimize DASD space and not write the end-time record. Use the
following guidelines for tuning this parameter.
• Use the default of PURGE=SPEED if you use the PURGE command to control your database.
• Set PURGE=DASD if you use KEEPSESS, because purge processing is not done and the end-time records
are not needed.
• Use PURGE=SPEED if you use both PURGE and KEEPSESS (using KEEPSESS selectively and using purge
processing for the rest of your database). PURGE=DASD has a more adverse effect on purge processing
than the extra end-time record has on session history recording.
• Set the PURGE parameter to DASD if you use the RESETDB command. There is no need for the end-time
record in this case. Ensure that your NLDM VSAM cluster definition (CNMSI101) has the REUSE operand.
The REUSE operand is required when using the RESETDB command.

SMDR Command
You can stop session monitor data recording by using the NLDM SMDR STOP command. Do not use this
command when a purge is running, because NLDM has separate control blocks for updating and deleting
records from the VSAM database in use. When the SMDR STOP command is issued, sessions on the VSAM
Record Queue go through normal cleanup processing, but the session history data is discarded instead

62 IBM Z NetView: Tuning Guide

 of being written to VSAM. Recording to the external log (SMF) proceeds normally and is not affected by
the SMDR command. The SMDR START command resumes the recording of session history data to VSAM.
Message AAU273I is sent to the NetView log stating that VSAM recording has resumed. This message also
notifies the user of the total number of sessions lost (session history data discarded) because of SMDR
STOP.
If a severe VSAM I/O error occurs, NLDM will automatically deactivate session monitor VSAM recording.
Browse the NetView log or system log for either message AAU272I or message AAU022I. Determine the
error by interpreting the major and minor return codes contained in the error message. When the VSAM
I/O error is resolved, issue SMDR START to resume VSAM recording.
SMDR QUERY can be used to determine database VSAM recording status. This is helpful if SESSMDIS is
showing a backup on the VSAM Record Queue.
See the IBM Z NetView Command Reference Volume 1 (A-N) or the NetView online help for a description of
the SMDR command.

SESSMDIS Command
The SESSMDIS command displays session monitor session counts, storage use, and workload traffic
information. SESSMDIS uses a view panel to display a subset of the counters that are written to the
external log with the NLDM RECORD STRGDATA command. Figure 13 on page 63 shows sample output
of the SESSMDIS command.

SESSMDIS Session Monitor Session and Storage Information 13:07:04 Page 1 of 1

1
Options in Effect SAW: YES LU Trace: NO CP/SSCP Trace: NO SESSTATS: NO
2
Session Counts CP-CP SSCP-SSCP SSCP-PU SSCP-LU LU-LU Filtered
Current: 0 0 1 35 8 0
Maximum: 0 0 1 35 8 0
3
Session Monitor Storage Usage
Resources: 8K Sessions: 16K Session Parms: 0K
PIU Trace: 0K SESSTATS: 0K RTM: 0K
RSCV: 0K Other: 1299K
Total: 1323K
4
VSAM Record Queue
Current: 0 Maximum: 3
5
Session Monitor Workload since 05/26/13 15:29:59
SAW Session Session PIU PIUs Sessions
Buffers Starts Ends Buffers Recorded
4 seconds: 0 0 0 0 0 0
Total: 86 89 44 0 0 44

ENTER= Refresh PF2= End PF3= Return

Figure 13. Session Monitor Session and Storage Information Panel

The following list describes the fields displayed:

1
Session monitor options in effect
SAW
Displays the setting of the SAW parameter in AAUPRMLP.
LU Trace
Displays the setting of the TRACELU parameter in AAUPRMLP.
CP/SSCP Trace
Displays the setting of the TRACESC parameter in AAUPRMLP.
SESSTATS
Displays the setting of the SESSTATS parameter in AAUPRMLP.

Chapter 6. Tuning for the Session Monitor 63

 2
Session counts (current and maximum)
CP-CP
The number of active CP-CP sessions for which the session monitor is maintaining session
awareness.
SSCP-SSCP
The number of active SSCP-SSCP sessions for which the session monitor is maintaining session
awareness.
SSCP-PU
The number of active SSCP-PU sessions for which the session monitor is maintaining session
awareness. This includes PU type 4 and PU type 2 sessions.
SSCP-LU
The number of active SSCP-LU sessions for which the session monitor is maintaining session
awareness.
LU-LU
The number of active LU-LU sessions for which the session monitor is maintaining session
awareness.
Filtered
The number of active sessions (of all types) that are being filtered by either the VTAM or the
NetView SAW filters.
3
Session monitor storage usage
Resources
The number of KB of storage allocated for resource information.
Sessions
The number of KB of storage allocated for session information.
Session Parms
The number of KB of storage allocated for session parameter information. This information is kept
only for sessions that are being PIU traced.
PIU Trace
The number of KB of storage allocated for PIU trace information. This storage is used for the PIU
wrap areas. Modifying the KEEPPIU parameter or the KEEPPIU option of the KCLASS statement
changes this number. The PIU wrap area for a session is allocated when the first PIU for the
session is received.
SESSTATS
The number of KB of storage allocated for session accounting or availability information.
• If accounting is active (SESSTATS=YES), the accounting storage for a session is allocated when
the first PIU for the session is received.
• If availability is active (SESSTATS=AVAIL), the availability storage for a session is allocated when
the start notification for the session is received.

RTM
The number of KB of storage allocated for response time monitor (RTM) data. Modifying the
KEEPRTM parameter can change this number. The RTM wrap area for a session is allocated when
the first RTM data is received for the session. If you do not collect RTM data with the COLLECT
command, the wrap area is allocated when the RTM data is received at session termination. A
small KEEPRTM parameter is useful in that case.
RSCV
The number of KB of storage allocated for route selection control vector data.

64 IBM Z NetView: Tuning Guide

 Other
This field contains miscellaneous storage being used by the session monitor, such as work storage
and internal control blocks.
Total
The total amount of storage allocated for active sessions. When sessions end, rather than freeing
the storage back to the operating system, the session monitor keeps the storage and reuses it
for new sessions (to avoid dynamic storage get and free requests). Therefore, the total storage
given on the SESSMDIS panel is not necessarily equal to the total virtual storage allocated by the
session monitor.

4
VSAM Record Queue (current and maximum)
The number of sessions that have ended and are waiting to be recorded to VSAM. This number
includes sessions that might have their session history recording filtered (see “DASD Option” on page
59). Sessions that have ended spend a few seconds on the VSAM record queue so that related data
can arrive from the network. If a large number of sessions end in a short amount of time, the record
queue can build up. By repeatedly refreshing the SESSMDIS panel, you can watch the queue shrink
and grow.
5
Session monitor workload traffic counters (total and 4 second rate)
The "since" time stamp indicates when the last NLDM RECORD STRGDATA command was issued or
when the session monitor was started. The following traffic counters represent the number of traffic
items processed since the time stamp.

SAW Buffers
The number of session awareness buffers that have been processed.
Session Starts
The number of session starts that have been processed.
Session Ends
The number of session ends that have been processed.
You can use the previous three numbers to estimate the SAW buffering ratio. The buffering ratio
can be calculated by summing the starts and ends, and dividing by the number of SAW buffers.
The SAW buffering ratio is useful in tuning the SAW buffer allocation. See “SAW Buffer Allocation
and Tuning” on page 50 for more information.

PIU Buffers
The number of PIU buffers that have been processed.
PIUs
The number of PIUs that have been processed.
You can use these numbers to estimate the PIU buffering ratio and the average fullness of the
PIU buffers. For the PIU buffering ratio, divide the number of PIUs by the number of PIU buffers.
See “PIU Buffer Allocation and Tuning” on page 54 for information on estimating the PIU buffer
fullness percentage and tuning the PIU buffer allocation.

Sessions Recorded
The number of sessions for which VSAM recording has taken place (sessions that are not DASD
filtered).
You can use the number of sessions with the Session Ends count to determine the NLDM session
recording rate. If the number of sessions recorded is equal to the number of session ends,
then no sessions have DASD recording filtered. If AAUTSKLP is showing high CPU usage from

Chapter 6. Tuning for the Session Monitor 65

 TASKUTIL output, consider filtering DASD recording. If the rate of session ends is constantly
high, determine if certain sessions are being restarted and immediately ending again. This can be
determined by running sample job CNMSJM10 and examining the output for specific sense codes
that would indicate a failure (for example: INITF, BINDF). See “DASD Filtering” on page 61 for
more information.
The SESSMDIS command can be run under an autotask and its output directed to the network log. You
can set a timer to run SESSMDIS under an autotask, so that you can examine the SESSMDIS output later
for workload activity trends. The total current explicit route (SART) count can be viewed. To obtain SART
data, the SESSMDIS command must be issued either in a window (for example, Window SESSMDIS) or
the results of the SESSMDIS command must be sent to the log. The SART data can not be displayed on
the SESSMDIS window because there is not enough space for this. An additional line of output is returned
for the Window SESSMDIS command and it is also returned to the NetView log. It appears at the end of
the SESSMDIS output in the following format:

TOTAL CURRENT EXPLICIT ROUTES (SARTS): 37

Use the SART information as a guide in setting the value of the NLDM.ERCOUNT statement in the
CNMSTUSR or CxxSTGEN member, or in the AAUPRMLP member. The complete Window SESSMDIS
output looks similar to the following example:

DSI378I SESSMDIS DISPLAY

SESSMDIS SESSION MONITOR SESSION AND STORAGE INFORMATION 13:14:26
OPTIONS IN EFFECT SAW: YES LU TRACE: NO CP/SSCP TRACE: NO SESSTATS: NO
SESSION COUNTS CP-CP SSCP-SSCP SSCP-PU SSCP-LU LU-LU FILTERED
CURRENT: 0 1 1 29 5 0
MAXIMUM: 0 1 1 29 6 0
SESSION MONITOR STORAGE USAGE
RESOURCES: 8K SESSIONS: 12K SESSION PARMS: 0K
PIU TRACE: 0K SESSTATS: 0K RTM: 0K
RSCV: 0K OTHER: 1673K
TOTAL: 1693K
VSAM RECORD QUEUE
CURRENT: 0 MAXIMUM: 2
SESSION MONITOR WORKLOAD SINCE 05/16/13 13:13:51
SAW SESSION SESSION PIU PIUS SESSIONS
BUFFERS STARTS ENDS BUFFERS RECORDED
4 SECONDS: 0 0 0 0 0 0
TOTAL: 4 39 2 0 0 2
TOTAL CURRENT EXPLICIT ROUTES (SARTS): 1

RTM Data Collection

The COLLECT RTM command gathers response-time data from PUs with the RTM feature.
If you collect data from a large network, issue separate COLLECT commands for subsets of the PUs rather
than collecting from all of the PUs at one time. If you use COLLECT *, you can cause a large burst of RUs to
flow in the network. Stagger the separate COLLECT commands to spread out the flow of RUs over a longer
period of time.

LUCOUNT Parameter
The LUCOUNT parameter is in the initialization member for the AAUTSKLP (default AAUPRMLP) member
and in the CNMSTYLE initialization member.
If you use the CNMSTYLE member, modify LUCOUNT in the CNMSTUSR or CxxSTGEN member instead of
in the AAUPRMLP member. The following example shows the default entry in the CNMSTYLE member:

NLDM.LUCOUNT=4000

Accurate use of the LUCOUNT parameter is pertinent for Session Monitor performance. The LUCOUNT
parameter provides the following functions:
• Specifies the number of LUs known by the session monitor. The default value is 4000.
• Optimizes control block search algorithms.

66 IBM Z NetView: Tuning Guide

• Allocates session-related storage.

Specify the number of LUs in your network for LUCOUNT. This number should include SNA interconnection
(SNI) sessions in your network. The number you specify does not have to be exact; however, too small
a value could hinder access to session blocks and cause an increase in NLDM initialization time and CPU
usage. A value in excess of the number of LUs can result in unused virtual storage but might improve
access to session blocks. It is better to overestimate LUCOUNT than underestimate. For every 250 LUs
specified, the search tables require 4K virtual storage.
Users must use the SESSMDIS command to determine a value for LUCOUNT. After the complete network
is activated, issue the SESSMDIS command and simply add the SSCP-LU and LU-LU session counts
(maximum, not current). This value provides a conservative estimate for the LUCOUNT. As most networks
are expanding, periodically monitor the coding of LUCOUNT as previously outlined. Having a LUCOUNT
that is coded too low is a common problem and impacts your performance.

Chapter 6. Tuning for the Session Monitor 67

68 IBM Z NetView: Tuning Guide
Chapter 7. Tuning for the NetView Management
Console

The NetView management console uses interactive graphics to display pictures, or views, that represent a
network, a portion of a network, or a group of networks at various levels of detail. These views show the
network resources and systems that you are monitoring. When you monitor a network, resource status
changes are reflected graphically in the views.
The NetView management console consists of a two tier client-server relationship.
Client and server code is installed on each workstation and runs as a Java™ application, displaying data
from RODM. The server portion of the workstation code, in turn, acts as a client requesting data from
the host NetView program. The NetView management console was adapted from the graphic data server,
therefore, most of the GMFHS code is unaffected.
Because the NetView management console function is accomplished through cooperative processing
between the host and the programmable workstation components, this section describes tuning
techniques for each component. For more information about host tuning techniques, see Chapter 7,
“Tuning for the NetView Management Console,” on page 69.

Workstation Tuning Techniques

This section describes tuning techniques for the NetView management console that can improve
performance at the workstation. These tuning techniques are arranged in order of expected effect on
performance, with the most important tuning considerations listed first.
1. Determine the amount of working set storage required for the NetView management console on the
workstation. See “Storage Estimates” on page 70.
2. Use powerful client workstations if there is a significant amount of view update activity and a large
total number of resources are displayed in the active views. See “Hardware Requirements” on page
70 and “Client Performance” on page 71.
3. Limit the background pictures used in client workstation views. See “Using Background Pictures” on
page 71.
4. Use high-speed connections instead of low-speed connections between the status focal point and the
server workstation. See “Status Focal Point to Programmable Workstation Connectivity” on page 71.
5. Take advantage of the server-client configuration features. See “Server-Client Configurations” on page
72.
If you are experiencing delays when views are being built on the workstation, consider the following;
• When running on any Windows workstation, open Task Manager and select the Performance window.
This window shows storage and system usages. Other platforms have similar tools. This can help
determine whether data is being sent or processed in the event of a potential suspended (hung)
situation.
• Be careful when clicking on a view. Watch the spinning icon on the top right of the NetView management
console Client window. When it is spinning, data is being requested/processed.
• Do not click on another view until the previously selected view is completely displayed. Doing so causes
the NetView management console to suspend or possibly even abend.
• If a view seems to be suspended and is not displayed, use the exit door to logoff of the NetView
management console Client. Check the client, server, and NetView logs for an indication of a potential
problem. If no error messages are found, attempt logging back on to the client.
• Whenever possible, try to not have background views open. Status changes are dynamic in the NetView
management console. If multiple views are open, each one will be checked to determine if it needs to
be updated. To close background views, right-click the background view and select the Close option.

© Copyright IBM Corp. 1997, 2022 69

 • Monitor all of the clients on each server. Make sure that clients logoff when not using the NetView
management console and at the end of each day. Failure to do this results in excessive network traffic
and potential network performance problems.
• Have only one component of the NetView management console code client or server active on a
workstation at a time to maintain stability and performance.
• If Task Manager on the client shows excessive CPU usage because of Java and the NetView
management console, consider upgrading the client workstations CPU.
• For improved performance when multiple views are active or when views are being cycled, run with a
minimum graphics resolution. To set the resolution, adjust the color setting in your display properties.
Use a 32-bit or lower resolution (such as 32-bit High Color or 16 bit Medium Color). Do not use a 64 bit
or higher resolution (such as 64 bit True Color).
In addition to the previous tuning considerations, several operands can affect the NetView management
console performance.
In the server.properties file that is included with the NetView management console, the following value
affects performance:
• statusUpdateAndViewChangeTicker — Default is 1000 (1 second).
• statusUpdateInterval — Default is 1.
• statusFlushCount — Default is 15.
• viewChangeInterval — Default is 3.
• viewChangeFlushCount — Default is 10.
These values should all be lowered simultaneously. The new values are picked up when the NetView
management console server is recycled. Monitor the NetView management console performance after
making the changes.

Storage Estimates
To ensure optimum performance of the workstation-based components of the NetView management
console, minimize the amount of paging and swapping that can occur. You can do this by providing as
much memory as will be used on a regular and consistent basis. There must be enough workstation
memory to consistently contain the working set. If severe memory constraints exist, abends can occur.

Hardware Requirements
This section lists hardware requirements for the following environments:
• “Intel Platform Workstations” on page 70
• “UNIX Platform Workstations” on page 70
• “Servers and Consoles” on page 70

Intel Platform Workstations

For Intel platform workstations, ensure that the minimum hardware requirements as outlined in the Z
NetView Program Directory are met.

UNIX Platform Workstations

For UNIX platform workstations, ensure that the minimum hardware requirements as outlined in the Z
NetView Program Directory are met.

Servers and Consoles

The following requirements are additional requirements beyond the minimum requirements for Intel or
UNIX platform workstations:

70 IBM Z NetView: Tuning Guide

 • 256 MB (RAM) for each console instance
• Between 256 and 384 MB of swap/page space
The formula for calculating each console instance is: 1.5 to 2.5 times the amount of RAM for systems
with 256 MB (or more).
Note:
1. If you have less than the minimum amount of RAM, you could easily have situations where you will
page heavily. Not only will this degrade performance, but you will probably require more page/swap
space for this activity. If this occurs, add another 33% to the combined total of memory and swap/
page space.
2. If you run other applications on these systems, increase the amount of installed RAM and page/swap
space accordingly to provide room for the additional applications to work without adversely affecting
the NetView management console.
3. For LINUX on z/Series requirements, refer to the README file that is included with the NetView
management console code.

Client Performance
If many resources are in the views displayed at a client workstation, consider using a high-performance
processor for this client workstation. The activity of drawing and redrawing the views can take
considerable amounts of processor resources. This resource use can be particularly high when view
navigation is necessary during the arrival and display of a large number of resource status updates. This
situation is likely when a large burst of status changes occurs, such as during network activation.
Ensure that the minimum hardware requirements, as outlined in the Z NetView Program Directory are met.
If the CPU and storage requirements are not met, then performance degradation will occur.
To minimize the number of resources actively displayed at a client workstation, use views containing
aggregate resources that represent the real resources you are monitoring. From aggregate resources, you
can navigate to views containing failing resources by using the fast path to failing resource feature. You
can customize the effect real resources have on an aggregate resource. Refer to the IBM Z NetView User's
Guide: NetView Management Console for information about view navigation and adjusting aggregation for
a resource.
Closing views that you are no longer using helps to ensure that client workstation resources are available
to handle a burst of client activity.
If you use the Cycle Windows window to cycle through views you have open for monitoring, lower values
for the cycle delay interval will result in higher view drawing activity.

Using Background Pictures

Adding a background picture to a view can impair client workstation performance. Including a background
picture can increase response times for the initial view display and whenever the view is redrawn. The
impaired performance is caused by the storage requirements of the background picture and the CPU
resources needed to draw it.
The CPU and memory requirements vary depending on the detail and complexity of the background
picture. If you are using detailed or complex background pictures, increased storage requirements can
increase response times for any other activity on the client workstation if these requirements cause your
working set storage requirements to exceed the amount of available memory.

Status Focal Point to Programmable Workstation Connectivity

The speed of the connections from the status focal point to the server workstation has a large impact on
NetView management console performance for large bursts of updates.
Processing of bursts generated by network failures is affected by link speed more than is processing of
bursts generated by network recovery.

Chapter 7. Tuning for the NetView Management Console 71

 These two factors combined cause the number of status changes per unit of time to be greater for
network failure than for recovery. Because of these factors, the larger throughput provided by higher-
speed connections is of more benefit during network failures.

Server-Client Configurations
Because the IP communication feature is being used as the communication vehicle for the NetView
management console, you can connect console workstations with the server workstation in a number
of ways. For example, you can have the server workstation attached to a token ring and in session with
console workstations on the same ring. The client workstation can be anywhere that an IP session can
be established. Also, all NetView management console workstations can be combined server-console
workstations.
When you select the workstation server-console configuration, several factors can affect the overall
performance. The server workstation maintains an in-storage database of the current network resource
status for opened views. The activity associated with resource status changes is both CPU and memory
intensive. The server workstation should be used as a standalone server workstation. Do not monitor
views that can be subject to continuous or large bursts of status update activity because updating the
graphical views is CPU intensive. Also, ensure that other applications that might be running on the server
and console workstation are not CPU and memory intensive.

Host Tuning Techniques

This section describes tuning techniques for the NetView management console at the host.

NETCONV
You can use either IP or LU 6.2 with the NETCONV command. If you are using LU 6.2, establish an LU
6.2 communication session between the status focal point host and server workstation. When this session
is established, the status focal point forwards the current status of all monitored resources to the server
workstation.
If the NETCONV command is issued before the network is activated, the focal point forwards the initial
status (never active) of all the monitored resources to the server workstation. When the network is
activated, the current status of all monitored resources is also forwarded to the workstation.
Issue the NETCONV command after network activation to avoid processing multiple status updates at
network activation time and reduce elapsed time for status display.
See the IBM Z NetView Command Reference Volume 1 (A-N) or the NetView online help for more
information about using the NETCONV command.

DUIGINIT Parameters
The following parameters in DUIGINIT (the GMFHS initialization member) affect the NetView
management console performance:
• LCON-STATUS-DELAY-MAX (The default value is 10.)
• LCON-STATUS-DELAY-TIME (The default value is 50.)
• LCON-EVCHANGE-BUFFER-INTERVAL (The default value is 500 [5 seconds]).
• LCON-AGG-BUNDLE-INTERVAL (The default is 500 [5 seconds]).
Use the following procedure to change the parameters:
• Reduce all parameters simultaneously
• Recycle GMFHS
• Monitor the NetView management console performance to determine if the changes have improved the
NetView management console performance.
Refer to the IBM Z NetView Administration Reference for more information about these parameters.

72 IBM Z NetView: Tuning Guide

Chapter 8. Tuning for the Resource Object Data
Manager

The NetView Resource Object Data Manager (RODM) is a data cache that is designed to store network
configuration and status information about system resources. You can use RODM to automate network
management functions associated with the resources defined to RODM. In addition, you can write RODM
applications to perform other network management and automation tasks.
For more information on the object-oriented terms used by the NetView program to describe RODM and
its data model, see the IBM Z NetView Resource Object Data Manager and GMFHS Programmer's Guide.

Tuning Techniques
The following major tuning techniques for RODM are described in this chapter:
1. Keep RODM logging to a minimum for production systems by using LOG_LEVEL 8. See “Customization
Parameters” on page 79.
2. See “Estimating Storage Usage” on page 115 to determine virtual storage use and DASD space
requirements for the checkpoint data sets. See “RODM Data Sets” on page 74. Ensure that the
MVS system has an adequate paging system. If your system is storage-constrained, consider moving
workloads to other systems or enhancing the MVS paging system.
3. Checkpoint RODM whenever there have been significant changes to the structure or topology of the
objects in RODM, such as after loading objects with the loader facility. Do not use checkpoints to
capture the status of objects in RODM. When a RODM warm start is performed, the RODM applications
which created the objects should update the object status when the application is initialized after the
RODM warm start. A warm start is relatively fast. See “Warm Start and CHKPT Commands” on page
73.
4. Generate RODM API statistics to analyze the content and activity of RODM. See “RODM API Statistics”
on page 74.
5. Generate RODM cell pool statistics to analyze the storage usage of RODM. See “RODM Cell Pool
Statistics” on page 76.
6. Specify a minimum number of CONCURRENT_USERS. Extra storage is required for each user. See
“Customization Parameters” on page 79.
7. Run RODM at the same dispatching priority as the NetView program.
8. If you are writing a RODM application, see “Programming Recommendations” on page 79.
9. The maximum number of objects supported is 2 097 135. For objects typical of the topology and
status monitoring of the NetView program, the operational number of objects is estimated to be
1 500 000. No code changes have been made to increase the number of classes and the number of
fields supported by RODM.
Note: To use the larger limit on the number of objects, you must cold start RODM. Otherwise, the
original limit is used if RODM is warm started. When RODM was cold started and a checkpoint was
taken, subsequent warm started instances of RODM continue to use the new maximum number of
objects.

Warm Start and CHKPT Commands

A warm start takes significantly less time than a cold start. The actual data still resides on the data-in-
virtual (DIV) checkpoint data set and is brought into extended storage (in 4 KB pages) as the data is
referenced. The warm start provides only an in-storage map of the DIV data set.
Take a checkpoint of RODM when there have been significant changes to the structure or topology of the
objects in RODM, such as after loading objects with the loader facility. The checkpoint provides a map of

© Copyright IBM Corp. 1997, 2022 73

 the latest changes to the VSAM linear data sets (LDSs). Do not use checkpoints to capture the status of
objects in RODM. When a RODM warm start is performed, the RODM applications that created the objects
should update the object status when the application is initialized.
Note: If you do not use the RODM checkpoint function, you can disable it and avoid allocating the
checkpoint data sets described in “RODM Data Sets” on page 74. To disable the checkpoint function,
comment out the DD statements for EKGMAST, EKGTRAN, and EKGD00X in your copy of the RODM
startup procedure (EKGXRODM).

RODM Data Sets

There are four data definition names in the supplied RODM JCL sample EKGXRODM that define the RODM
checkpoint VSAM linear data sets (LDSs). The data definition names are:
EKGMAST
The data set for the master window. Use the system defaults for the master window size, which is 4
cylinders.
EKGTRAN
This is the checkpoint data set for the segment window storage. The preferred allocation is 96
cylinders, which is the default.
EKGD001 and EKGD002
The checkpoint data sets for data window (checkpoint) storage. Allocate sufficient checkpoint storage
to contain all data model storage, plus extra storage in case your system has a burst of activity that
overruns the current allocation. The default space allocation is 72 cylinders.
See “Estimating Storage Usage” on page 115 to determine space allocations for these data sets. If you
are migrating from an earlier release of NetView, it is likely that you will need to increase your checkpoint
data set allocations.
The sample VSAM LDS definitions are in the EKGDWIND member.
You can use multiple checkpoint data sets to minimize the average size needed and provide for increased
efficiency during the checkpoint. You can define additional checkpoint LDSs by adding data definition
names in the EKGDWIND member (such as EKGD003 and EKGD004).
If you do not use the RODM checkpoint function, you can disable it and avoid allocating the checkpoint
data sets. To disable the checkpoint function, comment out the DD statements for EKGMAST, EKGTRAN,
and EKGD00X in your copy of the RODM startup procedure (EKGXRODM).
Error Messages:
1. If you receive error message EKG1110I, increase the size of the EKGTRAN data set for segment
window storage. Because you can define only one LDS for the segment window allocation, a cold start
is required.
2. If you receive error message EKG1111I, increase your space allocation for the EKGD001 and
EKGD002 data sets, or add VSAM checkpoint LDSs (such as EKGD003 and EKGD004).
• Error message EKG1111I is issued even though you are not taking a checkpoint, when the size of
your data model (classes and objects) grows beyond the size of the data windows present in your
checkpoint data sets, another data window cannot be allocated. RODM continues to operate until the
last data window is filled.
• You can take a checkpoint and warm start RODM when adding a checkpoint LDS.
• If the existing LDSs are reallocated with additional space, you must cold start RODM.

RODM API Statistics

RODM API statistics enable you to analyze the content and activity of RODM. Use the STATAPI parameter,
of the MVS MODIFY command to generate API statistics in a type 8 log record and the CLEAR option to
clear the counters.

74 IBM Z NetView: Tuning Guide

RODM API statistics give you the number of times a call was made or a method triggered since RODM
was cold started and since the counters were cleared using the MVS MODIFY rodmname,STATAPI,CLEAR
command for the following categories:
• User API (UAPI) calls
• Method API (MAPI) calls
• Object-specific (OS) methods triggered by RODM
• Object-independent (OI) methods triggered by users of RODM
RODM API statistics also specify the number of times a call was a success or failure. These statistics also
specify the type of method:
• Change
• Notify
• Query
Figure 14 on page 75 is an example of the output from the RODM log formatter for log record type 8 API
statistics.

Log_type : 8 (Statistics) RBA : 6213116

Record number : 31277 Record Length : 1324
Transaction ID: 0000000000000000x Timestamp : Sat Mar 30 11:33:30 2013
User Appl ID :
API Version : 1
Stat Type : 5 (API Statistics )
Last Clear Timestamp : Fri Mar 29 19:59:03 2013
Output Timestamp : Sat Mar 30 11:33:31 2013
No. of Query Triggered : 0
No. of Change Triggered : 18140
No. of Notify Triggered : 2273
No. of Objdel Triggered : 0
No. of Permanent Entries: 14
Permanent Count Data :
Function ID : 1302 (Create a Class)
Perm UAPI Count : 0000000000000044x
Perm MAPI Count : 000000000000001Ex
Function ID : 1304 (Create a Field)
Perm UAPI Count : 00000000000006F6x
Perm MAPI Count : 000000000000008Cx
Function ID : 1306 (Create a Subfield)
Perm UAPI Count : 0000000000001B67x
Perm MAPI Count : 0000000000000252x
Function ID : 1405 (Link 2 Objects - Methods Triggered)
Perm UAPI Count : 000000000000A091x
Perm MAPI Count : 0000000000005D0Ex
Function ID : 1407 (Unlink 2 Objects - Methods Triggered)
Perm UAPI Count : 0000000000000801x
Perm MAPI Count : 000000000000006Dx
Function ID : 1409 (Create an Object)
Perm UAPI Count : 00000000000030ACx
Perm MAPI Count : 0000000000000267x
Function ID : 1410 (Delete an Object)
Perm UAPI Count : 0000000000000126x
Function ID : 1412 (Add Notification Subscription)
Perm UAPI Count : 0000000000000003x
Perm MAPI Count : 000000000000147Fx
Function ID : 1413 (Delete Notification Subscription)
Perm MAPI Count : 00000000000004CCx
⋮

Figure 14. RODM Log Record Type 8 for API Statistics (Part 1 of 2)

Chapter 8. Tuning for the Resource Object Data Manager 75

 No. of Regular Entries : 48
Regular Count Data :
Function ID : 1101 (Connect to RODM)
Success UAPI Count: 10
Function ID : 1102 (Disconnect from RODM)
Success UAPI Count: 7
Function ID : 1302 (Create a Class)
Success UAPI Count: 68
Success MAPI Count: 30
Function ID : 1304 (Create a Field)
Success UAPI Count: 1782
Success MAPI Count: 140
Function ID : 1306 (Create a Subfield)
Success UAPI Count: 7015
Success MAPI Count: 594
Function ID : 1401 (Change a Field)
Success UAPI Count: 111544
Fail UAPI Count : 12423
Success MAPI Count: 182551
Function ID : 1403 (Change a Subfield)
Success UAPI Count: 102
Success MAPI Count: 18151
Function ID : 1405 (Link 2 Objects - Methods Triggered)
Success UAPI Count: 41105
Fail UAPI Count : 9931
Success MAPI Count: 23822
Fail MAPI Count : 7079
Function ID : 1407 (Unlink 2 Objects - Methods Triggered)
Success UAPI Count: 2049
Success MAPI Count: 109
Function ID : 1409 (Create an Object)
Success UAPI Count: 12460
Fail UAPI Count : 23
Success MAPI Count: 615
Function ID : 1410 (Delete an Object)
Success UAPI Count: 294
Fail UAPI Count : 2

Figure 15. RODM Log Record Type 8 for API Statistics (Part 2 of 2)

If you specify the CLEAR option when writing the API statistics, the regular count data counters are reset
to zero after being written to the log (the permanent count data counters are not affected by the CLEAR
option). You might find it convenient to write the API statistics on a timer basis using the NetView EVERY
command. If you use the CLEAR option to clear the counters each time you write them, the counters show
which API calls were made during the timer interval. Clearing the counters periodically also ensures that
the counters do not overflow.
See the IBM Z NetView Troubleshooting Guide for a description of all of the fields in RODM log record type
8.

RODM Cell Pool Statistics

You can use RODM cell pool statistics to determine whether RODM is using storage efficiently. To
access these statistics, use the MVS MODIFY command with the STATCELL parameter (MVS MODIFY
rodmname,STATCELL) to write cell pool use information to a type 8 RODM log record. See the IBM Z
NetView Troubleshooting Guide for a description of all of the fields in RODM log record type 8.
Figure 16 on page 77 shows an example of the output from the RODM log formatter for log record type 8
segment and window statistics.

76 IBM Z NetView: Tuning Guide

 Log_type : 8 (Statistics) RBA : 6211080
Record number : 31276 Record Length : 2036
Transaction ID: 0000000000000000x Timestamp : Sat Mar 30 11:33:31 2013
User Appl ID :
API Version : 1
Stat Type : 1 (Window Statistics)
Current pocket: 1
Avail. pocket : 1
No. of Entries: 33
Cell Size ( 0): 8 Pool Size : 1
No. in Use : 154233 High Water Mrk: 156185
In Use Percent: 20 Total Inuse % : 20
High Water % : 20
Histogram Data :
( 0) 0 ( 1) 0 ( 2) 0 ( 3) 0
( 4) 0 ( 5) 78448 ( 6) 19212 ( 7) 69835
Cell Size ( 1): 12 Pool Size : 1
No. in Use : 167074 High Water Mrk: 167074
In Use Percent: 22 Total Inuse % : 22
High Water % : 22
Histogram Data :
( 0) 0 ( 1) 1147 ( 2) 0 ( 3) 8285
( 4) 0 ( 5) 0 ( 6) 0 ( 7) 161887
Cell Size ( 2): 16 Pool Size : 1
No. in Use : 35885 High Water Mrk: 35885
In Use Percent: 4 Total Inuse % : 4
High Water % : 4
Histogram Data :
( 0) 0 ( 1) 964 ( 2) 0 ( 3) 16281
( 4) 0 ( 5) 759 ( 6) 0 ( 7) 26268
Cell Size ( 3): 24 Pool Size : 1
No. in Use : 27278 High Water Mrk: 27279
In Use Percent: 3 Total Inuse % : 3
High Water % : 3
Histogram Data :
( 0) 8 ( 1) 41631 ( 2) 1844 ( 3) 7
( 4) 110 ( 5) 11 ( 6) 768 ( 7) 1048
Cell Size ( 4): 32 Pool Size : 1
No. in Use : 1801 High Water Mrk: 1802
In Use Percent: 0 Total Inuse % : 0
High Water % : 0
Histogram Data :
( 0) 12 ( 1) 43 ( 2) 130 ( 3) 15305
( 4) 14 ( 5) 4 ( 6) 691 ( 7) 14015
⋮

Figure 16. RODM Log Record Type 8 for Segment and Window Statistics

The following describe the segment and window statistic fields in log record type 8, which is shown in
Figure 16 on page 77.
NO. OF ENTRIES
Specifies the number of entries in the cell pool array.
CELL SIZE
Specifies the cell size in bytes as defined in member EKGCUST.
POOL SIZE
Specifies the number of 4 KB pages that are allocated when a pool extension is needed (defined in
member EKGCUST).
NO. IN USE
Specifies the number of cells that are unavailable.
HIGH WATER MRK
Specifies the high-water mark for in-use cells.
IN USE PERCENT
Specifies the percentage of in-use cells.
TOTAL INUSE %
Specifies the percentage of total cells in use.
HIGH-WATER %
Specifies the percentage for the high-water mark.

Chapter 8. Tuning for the Resource Object Data Manager 77

 HISTOGRAM DATA
Lists eight counters associated with the cell size. These counters are used to show the distribution of
storage size requests satisfied by the cell pool.
To evaluate the amount of window storage currently in use, multiply the number of cells in use (NO. IN
USE) by the cell size for each of the entries in the cell pool array. The sum of the products is the amount of
RODM data space storage currently in use.

Using the Histogram Data

Provisions for monitoring the allocation requests and the usage of data space storage are built into the
RODM storage manager. As each request for storage is received and matched to a cell, an additional
calculation is made. A table is allocated at initialization that contains the following information for each
cell pool defined to the system:
• The cell size in bytes
• The pool size in pages
• The number of cells available
• The number of cells in use
• The high-water mark of cells in use
• A set of eight counters to reflect the approximate size of the actual storage request (plus 4 bytes of
control information) in histogram format
Figure 17 on page 78 is an example of histogram data.

Cell Size ( 6): 48 Pool Size : 1

⋮
Cell Size ( 7): 64 Pool Size : 1
No. in Use : 4817 High Water Mrk: 4817
In Use Percent: 2 Total Inuse % : 2
High Water % : 2
Histogram Data :
( 0) 0 ( 1) 0 ( 2) 0 ( 3) 0
( 4) 6 ( 5) 4813 ( 6) 0 ( 7) 0

Figure 17. Histogram Data

The histogram data can be used to tune the customizable cell pool sizes. To evaluate the histogram data
for Figure 17 on page 78:
1. Subtract the previous cell size from the current cell size (64 - 48 = 16). This is the size of the range of
storage requests that are serviced by this cell pool.
2. Divide the result by 8 (16 / 8 = 2). This is the size of the range of storage requests for each of the eight
counters in the histogram data.
3. Add 1 to the position value (0–7 becomes 1–8).
4. Multiply each position in the histogram by the result from “2” on page 78:

1×2=2
2×2=4
3×2=6
4×2=8
5 × 2 = 10
6 × 2 = 12
7 × 2 = 14
8 × 2 = 16
5. Add these results to the previous cell size (48) to get the maximum storage request size counted in
each histogram position:

78 IBM Z NetView: Tuning Guide

 Position (0) (1) (2) (3) (4) (5) (6) (7)
Increment 2 4 6 8 10 12 14 16
Size 50 52 54 56 58 60 62 64

The following results are from Figure 17 on page 78:

Size Requested
Number of Requests
Greater than 48 but less than or equal to 50
0
Greater than 50 but less than or equal to 52
0
Greater than 52 but less than or equal to 54
0
Greater than 54 but less than or equal to 56
0
Greater than 56 but less than or equal to 58
6
Greater than 58 but less than or equal to 60
4813
Greater than 60 but less than or equal to 62
0
Greater than 62 but less than or equal to 64
0
The results show that 4813 requests were for storage greater than 58 bytes but less than or equal to 60
bytes (59 or 60 bytes). If a cell size of 60 had been defined to RODM, 4 bytes times 4813 requests, or
19252 bytes, would have been saved.
Any additions or modifications you make to the cell pool sizes should be specified in the CELL_POOL
definitions in EKGCUST. All cell sizes must be full words. Consider how many cells fit on a page, because
all pool sizes are rounded up to the next page boundary.

Customization Parameters
The defaults supplied with the NetView program in EKGCUST are adequate for most systems. However,
you can alter the following settings to meet your specific requirements.
• The number of CONCURRENT_USERS is initially set to 10. You might need to increase this value, but do
not make it unnecessarily high because extra storage is required for each user.
• The number of ASYNC_TASKS is initially set to 5. To save storage, you can decrease this value to 2.
• LOG_LEVEL is initially set to 8 (record errors only). This is preferred for a production environment where
more than one RODM API call a second is anticipated. If the log levels were changed on a test system
for debugging purposes, consider changing LOG_LEVEL and MLOG_LEVEL back to 8 when switching to
production.
• Method tracing should be done only for problem solving, not during production. Set MTRACE_TYPE to
X'00000000' to disable method tracing.

Programming Recommendations
This section contains programming recommendations for those writing RODM applications.
• Keep the number of RODM API calls to a minimum.
• When possible, use the following to combine multiple operations into a single API call:

Chapter 8. Tuning for the Resource Object Data Manager 79

 – Query multiple subfields (function ID 1508)
– Change multiple subfields (function ID 1419)
– Execute a list of functions (function ID 1600)
• Do not specify initial values on loader field definitions if they are not required. These values require
extra processing at load time.
• Query fields by ID rather than by name when possible.
• Consider user methods similar to online CICS or IMS transactions. Use a minimum of processing and
MAPI calls for methods that are likely to be run frequently.

80 IBM Z NetView: Tuning Guide

Chapter 9. Tuning for VSAM

Input/output (I/O), specifically direct access storage device (DASD) I/O, is a major concern of
performance and tuning, especially in a NetView environment. For databases, the NetView program uses
VSAM data sets. The NetView program records messages to a log data set. The NetView program puts
session data into a session monitor database, and network events and statistics into a hardware monitor
database.

Tuning Techniques
Following are the major VSAM tuning techniques, arranged in order of expected effect on performance,
with the most important tuning considerations listed first. These are described in detail in this chapter.
1. Use the CISIZE values that are used in the sample cluster definitions for best performance. Do not use
the same CISIZE values for the hardware monitor and session monitor databases.
2. Start with the default LSR buffer pool allocations, and monitor the buffer miss percentage with the
VSAMPOOL command. Increase the number of buffers for individual pools where needed, and reduce
the number of buffers for pools that are not used frequently. See “Local Shared Resources (LSR) and
Deferred Write (DFR)” on page 81 and “VSAMPOOL Command” on page 87.
3. Use the DBAUTO command to reorganize databases that are not deleted and redefined regularly. See
“ VSAM Database Maintenance” on page 88.
4. Consider using the deferred write (DFR) performance option for the hardware monitor facility database
to reduce I/O activity. The sample definitions for the session monitor database have DFR specified
already. See “Local Shared Resources (LSR) and Deferred Write (DFR)” on page 81.
5. Monitor VSAM database performance using the LISTCAT and VSAMPOOL commands. See “Monitoring
VSAM Performance” on page 84.

Local Shared Resources (LSR) and Deferred Write (DFR)

The VSAM performance options of local shared resources (LSR) and deferred write (DFR) provide major
improvements in VSAM database processing. LSR enables the sharing of common control blocks (I/O
control blocks, I/O buffers, and channel programs). On GET requests, buffers are searched for direct
record retrievals. Without LSR, VSAM performs I/O for direct retrievals regardless of whether the control
interval (CI) containing the desired record is in storage.
DFR causes VSAM to defer the write I/O when records are directly inserted or replaced in direct mode.
Without DFR, VSAM does not defer the I/O for direct inserts or replacement of records. With DFR, the
buffers are written in these instances:
• No more buffers are available to perform a retrieve.
• The application issues the WRTBFR macro indicating that VSAM should write out the modified buffers.
• The database is closed.

If the NetView program ends without closing the databases, the records in the DFR buffers are not written
to the databases.
If you specify DFR, you get both the LSR and DFR options.
Do not cancel the NetView program except as a last resort. If you must issue a FORCE command, try
to close the databases by issuing the NetView SWITCH command with the T option. This closes the
active database and does not perform a switch. If this procedure does not work, issue the NetView STOP
FORCE command for each active VSAM task. If you must use the MVS FORCE command to bring down
the NetView program and you have specified DFR, you might have to delete and redefine the affected
databases. The exposure of having records not written to the databases is minimized by the extended

© Copyright IBM Corp. 1997, 2022 81

 specify task abnormal exits (ESTAEs) that trap abends and close the databases. However, if the system
operator ends the NetView program with the MVS FORCE command, the ESTAEs are not driven.

Definitions for LSR and DFR

To define LSR and DFR values, code the DSTINIT statement shown in the following example in the
NetView initialization member for each data services task that uses VSAM.

DSTINIT MACRF=xxx

In this example, xxx is either LSR or DFR.

Table 2 on page 82 lists the MACRF and CISIZE values for the NetView components and facilities that
appear in the sample definitions.

Table 2. Sample MACRF and CISIZE Values for NetView Components

Component Member MACRF 3390 DASD
Index CISIZE Data CISIZE
Central site control DSIKINIT LSR 2048 7168
facility
Hardware monitor CNMSTYLE LSR 2560 18432
Network log n/a 1024 4096
Save/Restore DSISVRTD LSR 4096 8192
Session monitor AAUPRMLP DFR 1536 24576
TCP connection CNMSTYLE LSR 3584 26624
management
Trace log DSITRCBK LSR 512 16384

Note: The CISIZE values specified for the data and index components in the samples that are included
with the NetView program are based on using an IBM 3390 (using ICF catalogs). Because of the higher
capacity of the 3390, different data buffer selections were made. These new selections result in a new
index control interval size for one cluster. If other types of devices are used to allocate these clusters,
these operands might need to be adjusted for optimal use of the device.
LSR is the default for the hardware monitor database. Consider using DFR for these databases, because
DFR gives better performance. LSR is used in the samples because some environments cannot tolerate
losing records in the hardware monitor database, even though the possibility is remote.
Do not use LSR or DFR for the network log. The DSILOG task buffers records before writing them to DASD.
Log browse does not work if LSR or DFR is used for the network log.

Buffer Pool Sizes

The VSAM buffer pool allocation is defined in module DSIZVLSR. The VSAM BLDVRP (build VSAM resource
pool) macro creates the DSIZVLSR module. Figure 18 on page 83 shows the sample buffer pool
allocations for MVS sample member CNMSJM01.

82 IBM Z NetView: Tuning Guide

 DSIZVLSR CSECT
BLDVRP *
BUFFERS=(7168(4), DSIKREM *
8192(20), DSISVRT *
16384(4), DSITRACE *
18432(20), BNJDSERV *
22528(20), DSITCONT (old/migration) *
24576(20), AAUTSKLP *
26624(20)), DSITCONT *
KEYLEN=96, *
MF=L, *
MODE=24, *
RMODE31=BUFF, *
SHRPOOL=0, *
STRNO=40, *
TYPE=(LSR,DATA)
***********************************************************************
* NOTE:- DO NOT ADD ANY CODE BETWEEN THE BLDVRP DEFINITIONS. THIS MAY *
* CAUSE UNPREDICTABLE RESULTS IN NETVIEW PROCESSING. *
***********************************************************************
BLDVRP *
BUFFERS=(512(3), DSITRACE *
1536(30), AAUTSKLP *
2048(30), DSIKREM *
2560(30), BNJDSERV *
3584(30), DSITCONT *
4096(30)), DSISVRT *
MF=L, *
MODE=24, *
RMODE31=BUFF, *
SHRPOOL=0, *
TYPE=(LSR,INDEX)
END

Figure 18. Sample BLDVRP Macros Defining VSAM Buffer Pools for CNMSJM01

Note:
The buffer sizes in DSIZVLSR correspond to the default index and data CISIZE values, which are based on
using 3390 DASD (using ICF catalogs).
If there are buffer sizes defined in DSIZVLSR that none of your databases use, remove them to decrease
storage usage. Use the VSAMPOOL command to monitor usage of the LSR buffer pools. See “VSAMPOOL
Command” on page 87.

KEYLEN Parameter
The KEYLEN parameter specifies the maximum key length of the data sets that share this pool. This
keyword should be specified only on the data pool.

STRNO Parameter
The STRNO parameter specifies the potential number of requests, in the range of 1–255, that can be
issued concurrently for all the data sets sharing the resource pool. Set STRNO to the total of all the
DSRBO values for data services tasks that use the LSR resource pool. The sample value of 40 is sufficient
for most environments. If you modify the DSRBO values for some of the DSTs, ensure that you adjust the
STRNO parameter accordingly.

BUFFERS Parameter
The BUFFERS parameter specifies the size and number of buffers in each buffer pool in the resource
pool. When you open a database and specify LSR or DFR, VSAM looks for a buffer pool for the INDEX and
DATA components, depending on their control interval sizes. A buffer pool that is the same size as the
control interval is chosen. If a buffer pool with the same size has not been defined, the next higher buffer
pool size is chosen. Databases with the same control interval sizes share the same buffer pool. Allocate
enough buffers of a particular size to satisfy all DSTs sharing the buffer pool.

Chapter 9. Tuning for VSAM 83

 The BLDVRP macros are specified with values that separate the index and data control intervals into
separate pools. Having separate index and data pools allows the critical index records to remain resident
in memory without the need to allocate an excessive number of buffers.
The size of the LSR buffer pool allocation affects VSAM performance considerably. If you want to change
the buffer pool allocation, you can do so by modifying the sample definition and running the VSAM
BLDVRP macro to create a new DSIZVLSR module. Refer to IBM Z NetView Installation: Configuring
Additional Components for more information about running the VSAM BLDVRP macro.

Buffer Pool Size Recommendations

Consider the following tuning recommendations in determining the buffer pool sizes for your
environment.
• Start with the default number of buffers for each buffer size in DSIZVLSR. See Figure 18 on page 83.
• Monitor the buffer usage with the VSAMPOOL command.
– Remove buffer sizes that are never used to save storage.
– For buffer sizes that are used infrequently, consider reducing the number of buffers to save storage.
– For index buffers that are used frequently, the number of buffer finds (BFRFND) should be at least
10 to 20 times the number of buffer reads (BUFRDS). If this is not the case, consider increasing the
number of buffers to reduce I/O activity.
– For data buffers that are used frequently, the number of buffer finds should be greater than the
number of buffer reads. If this is not the case, consider increasing the number of buffers.
– When modifying the buffer allocations, monitor the buffer usage before and after making changes.
If an increase does not result in improved performance, reduce the buffer allocation to its previous
value.
See “VSAMPOOL Command” on page 87 for information about the command and for more information
on tuning LSR buffer allocations.

Allocating the Buffer Pools in an MVS Hyperspace

For MVS systems, you can allocate the VSAM buffer pools in a hyperspace*. Specifying hyperspace buffers
is a method of reducing I/O to DASD by caching data in expanded storage. When you access the buffer
pool, the page must first be moved from expanded storage to central storage. The alternative is to let the
operating system storage manager determine the location of pages using its page replacement strategy.
If your system central storage is constrained, allocating the buffer pool in a hyperspace might help relieve
this contention.
To use hyperspace buffers, the buffer sizes must be in multiples of 4096. Therefore, change the
corresponding CISIZE values for your databases to be multiples of 4096. Round the sizes up to the
nearest multiple of 4096. For example, if the data CISIZE of a database is 22528, increase the CISIZE to
24576.
Note: *Specifying hyperspace buffers requires a modification to the VSAM BLDVRP macro (which is used
to create module DSIZVLSR). See the appropriate MVS publication for a description of the syntax of the
BLDVRP macro.

Monitoring VSAM Performance

NetView provides the LISTCAT and VSAMPOOL commands to assist in assessing the performance of
VSAM. VSAMPOOL is available on MVS systems only.

LISTCAT Command
The LISTCAT command displays VSAM database definition and performance data for NetView data
services tasks that have open VSAM databases. The information is similar to the data from the access

84 IBM Z NetView: Tuning Guide

methods services (AMS) LISTCAT command; however, the NetView LISTCAT command provides the
information online, while the VSAM database is active.
The LISTCAT command is useful in tuning the VSAM databases and in validating the database definitions.
This command is a full-screen command processor. After invoking the command, press the ENTER key
each time you want updated information. The screens are automatically copied to the network log. If
LISTCAT is run on the primary program operator interface task (PPT) or an autotask, the information is
sent to the network log and command execution ends. This enables LISTCAT to run from a NetView timer
command. Figure 19 on page 85 shows a sample of output from the LISTCAT command. The IBM Z
NetView Command Reference Volume 1 (A-N) provides a brief description for each field that is displayed.

LISTCAT Listcat of Active VSAM Data Base for BNJDSERV 11:28:07

VSAM ACB Options: LSR, DFR, ADR, KEY, SEQ, DIR, OUT 1
Cluster Information: 2
DDNAME: BNJLGPR KEYLEN: ..........76 RKP: ...........0
BSTRNO: ...........0 STRNO: ..........11 STRMAX: ...........2
BUFSP: ...........0
DATA Component Information: 3
LRECL: ........4086 CINV: .......18432
BUFND: ..........12 BUFNO: ...........0
NEXT: ...........2 FS: ...........8
NCIS: ........1130 NSSS: ..........49
NEXCP: ........8588 NLOGR: .......60326 NRETR: .......21326
NINSR: .......70595 NUPDR: .......10569 NDELR: .......10278
AVSPAC: ....12390400 ENDRBA: ....28672000 HALCRBA: ....29245440
INDEX Component Information: 4
LRECL: ........2553 CINV: ........2560
BUFNI: ...........0 BUFNO: ...........0
NEXT: ...........3 NIXL: ...........2
NEXCP: ........2214 NLOGR: ..........51
AVSPAC: ........2560 ENDRBA: ......130560 HALCRBA: ......133120

Figure 19. Sample LISTCAT Command Output Using 3390 DASD

The following fields are useful in tuning the VSAM database.

1
VSAM ACB Options
NSR
No LSR and no DFR
LSR
Local shared resources
DFR
Local shared resources and deferred write
Under VSAM ACB options, the performance options of LSR and DFR are shown if they are in use for the
data set.
2
Cluster Information
STRNO
Number of VSAM strings currently active
STRMAX
Maximum number of strings used
These fields show the VSAM string activity for the data set. The number of VSAM strings (STRNO)
is defined on the VSAM BLDVRP macro. See “Definitions for LSR and DFR” on page 82 for more
information about the BLDVRP macro. The correct value for STRNO is the sum of the number of
DSRBOs for DSTs using the resource pool. Although no tuning is necessary here, these fields show
how the high-water mark for concurrent operations (STRMAX) compares to the number of VSAM
strings defined (STRNO).
3
Data Component Information

Chapter 9. Tuning for VSAM 85

 NEXT
The number of extents in the data component.
NCIS
The number of data control interval splits.
NSSS
The number of data control area splits.
NEXCP
The number of EXCP (execute channel program - SVC 0) macros issued by VSAM against the data
component.
NLOGR
The current number of records in the data component.
NRETR
The number of records that have been retrieved from the data component, whether they are for
updates.
NINSR
The number of records that have been inserted into the data component before the last record.
Records originally loaded and records added to the end are not included in this statistic.
NUPDR
The number of records that have been retrieved for update and rewritten. This value does not
reflect records that are deleted only, but a record that is updated and then deleted is counted.
NDELR
The number of records that have been deleted from the data component.
AVSPAC
The number of bytes available in the data component.
ENDRBA
The number of bytes used in the data component.
HALCRBA
The number of bytes allocated in the data component.
4
Index Component Information
NEXT
The number of extents in the index component.
NIXL
The number of levels of records in the index.
NEXCP
The number of EXCP (execute channel program - SVC 0) macros issued by VSAM against the index
component.
NLOGR
The current number of records in the index component.
AVSPAC
The number of bytes available in the index component.
ENDRBA
The number of bytes used in the index component.
HALCRBA
The number of bytes allocated in the index component.
Notes:
1. The logical record counts (NRETR, NINSR, NUPDR, and NDELR) show the cumulative number of total
records retrieved, inserted, updated, and deleted since the file was last created, indicating the type of
data set activity that has taken place.

86 IBM Z NetView: Tuning Guide

 2. If the number of levels in the index component (NIXL) is excessive, VSAM performance suffers.
Reorganize the data set so that the number of index levels is three or fewer.

VSAMPOOL Command
The VSAMPOOL command displays statistics about NetView VSAM resource pool utilization when the
NetView program has been defined to use local shared resources (LSR) or deferred writing of records
(DFR). The LSR resource pool is subdivided into buffer pools determined by control interval sizes. You
define the LSR resource pool and buffer pools with the DSIZVLSR module. See “Definitions for LSR and
DFR” on page 82 for more information.
VSAMPOOL lists all of the buffer pools that are using LSR and DFR. The output shows the total usage
per control interval size (CINV). The display shows separate statistics for the DATA and INDEX VSAM
LSR/DFR buffers that were defined in DSIZVLSR. Figure 20 on page 87 shows a sample of output from
the VSAMPOOL command.

CNM260I VSAM LSR/DFR RESOURCE POOL STATISTICS

BNH091I BUFFER TYPE = DATA
CNM948I CINV BUFNO BFRFND BUFRDS NUIW UIW ERCT
CNM261I 7168 4 0 0 0 0 0
CNM261I 8192 20 2 1 0 0 0
CNM261I 16384 4 0 0 0 0 0
CNM261I 18432 20 0 0 0 0 0
CNM261I 20480 20 0 0 0 0 0
CNM261I 22528 20 0 0 0 0 0
CNM261I 24576 20 0 0 0 0 0
BNH090I BUFFER TYPE = INDEX
CNM948I CINV BUFNO BFRFND BUFRDS NUIW UIW ERCT
CNM261I 512 3 0 0 0 0 0
CNM261I 1536 30 0 0 0 0 0
CNM261I 2048 30 0 0 0 0 0
CNM261I 2560 30 0 0 0 0 0
CNM261I 3072 10 0 0 0 0 0
CNM261I 3584 30 0 0 0 0 0
CNM261I 4096 30 2 1 0 0 0
CNM262I END OF DISPLAY

Figure 20. Sample Output from the VSAMPOOL Command Using 3390 DASD

The following information is displayed for each buffer pool (see Figure 20 on page 87).
CINV
Control interval size (or buffer size) for the buffer pool
BUFNO
The number of buffers in the buffer pool
BFRFND
The number of requests for retrieval that could be satisfied without an I/O operation (the data was
found in a buffer)
BUFRDS
The number of reads to bring data into a buffer
NUIW
The number of non-user initiated writes (writes that VSAM was forced to perform because no buffers
were available for reading the contents of a control interval)
UIW
The number of user-initiated writes (PUTs not deferred or WRTBFRs)
ERCT
The number of write errors that have occurred

The VSAMPOOL command is useful in tuning the size of the LSR buffer pools.
The most useful statistic is the lookaside hit ratio, which is calculated as follows:

Chapter 9. Tuning for VSAM 87

 BFRFND
lookaside hit ratio = ------------
BUFRDS

The lookaside hit ratio gives an indication of the adequacy of the LSR buffer allocation. The optimal
lookaside hit ratio depends on your environment. Consider the following in general:
• For data buffers, the lookaside hit ratio should be 1 or greater; values higher than 5 are unusual for
databases with high activity.
• For index buffers, the lookaside hit ratio should be 10 or greater. The number of buffer finds (BFRFND)
should be at least 10 times greater than the number of buffer reads (BUFRDS). The higher the lookaside
hit ratio, the better.

The following approach is suggested for tuning the allocation for the LSR buffer pools:
• Remove buffer sizes that are not used to save storage.
• For buffer sizes that are used infrequently, consider reducing the number of buffers to save storage.
• For frequently used buffer sizes, increase the number of buffers if the lookaside hit ratio is less than 10
for index buffers, or less than 1 for data buffers.
• When increasing the buffer allocation:
1. Monitor the lookaside hit ratio for the current buffer pool allocation.
2. Increase the number of buffers in one or more of the buffer pools. The new buffer allocation does not
take effect until the NetView program is stopped and restarted.
3. Monitor the lookaside hit ratio for the new allocation.
4. Repeat steps 2 and 3 until the lookaside hit ratio does not improve with an increase in the number of
buffers for the buffer pool.
• When decreasing the buffer allocation:
1. Monitor the lookaside hit ratio for the current buffer pool allocation.
2. Decrease the number of buffers in one or more of the buffer pools. (The new buffer allocation does
not take effect until the NetView program is stopped and restarted.)
3. Monitor the lookaside hit ratio for the new allocation.
4. Repeat steps 2 and 3 until the lookaside hit ratio degrades noticeably with a decrease in the number
of buffers for the buffer pool.
Note: You must monitor the VSAM buffer pool usage using the VSAMPOOL command. This command
determines whether the initial VSAM buffer allocations are sufficient. It is also a simple tool for
monitoring buffer usage after any changes are made. Run the VSAMPOOL command on a timer every
hour when changes are made, and use the procedure previously outlined to verify that performance is
acceptable.

VSAM Database Maintenance

Use the VSAM access method services (AMS) REPRO or IMPORT and EXPORT commands to reorganize
a database to recover space lost through CI and CA splits. Use of these commands can improve
performance in accessing the database and in reducing database size. Free space decreases the
likelihood of CI and CA splits thereby, improving performance. This, in turn, decreases the chance that
VSAM will move a set of records to a different cylinder away from other records in the key sequence.
When a direct insert occurs that does not result in a split, VSAM inserts the records into available free
space.
When VSAM is called to delete or replace a record, it flags the space as deleted and writes a new record.
However, the deleted or replaced record is not actually deleted, but is only flagged as deleted, and
continues to consume space. The accumulation of these flagged records results in less available free
space within a VSAM data cluster. A query against a VTOC might indicate that free space is available on
a disk pack, but this does not mean that empty control intervals (CI) are available within the database

88 IBM Z NetView: Tuning Guide

cluster for new Data or Indexes records. To delete these flagged records completely, periodically run the
REORG command on all VSAM clusters. To view the amount of free space that is available in a KSDS
cluster, run a LISTCAT on the Cluster, then look at the values for FREESPC-BYTES in any DATA and INDEX
sections. To get the number of completely empty control intervals that are still available to hold new
records, divide the FREESPC-BYTES value by the CISIZE value. To get the approximate number of records
that can still be written, divide the FREESPC-BYTES value by the record size (PHYREC-SIZE).
You can perform NetView VSAM database maintenance from the NetView program without having to shut
down the VSAM database. Use the NetView IDCAMS command to run the AMS utility commands that are
stored in a data set.
You can use the PURGEDB command to delete unneeded data from the hardware monitor and session
monitor VSAM databases. If you use PURGEDB, use the VSAM AMS REPRO or IMPORT and EXPORT
commands to reclaim free space.
To delete historical data from the hardware monitor or session monitor databases, use the RESETDB
command to clear the databases when the NetView program is active. The RESETDB command is simpler
than deleting and redefining the databases and is faster than using the PURGEDB command to purge the
entire database. Ensure sure that the REUSE operand is coded on the session monitor cluster definition
(CNMSI101). The RESETDB command requires the REUSE parameter.
Use the DBAUTO command to perform maintenance on your VSAM databases while the NetView program
is running. The DBAUTO command works on the session monitor, hardware monitor, and Save/Restore
databases. You can perform the following operations:
• Switch to a secondary database
• Selectively purge entries in the session monitor or hardware monitor databases
• Reorganize a database
• Erase the contents of a database
See the IBM Z NetView Command Reference Volume 1 (A-N) or the NetView online help for a description of
the syntax and use of the IDCAMS, RESETDB, DBAUTO, and PURGEDB commands.

Chapter 9. Tuning for VSAM 89

90 IBM Z NetView: Tuning Guide
Chapter 10. Additional Tuning Considerations

This section provides miscellaneous tuning information.

Tuning Considerations
The tuning considerations are arranged in order of expected effect on performance, with the most
important listed first. These, along with other performance considerations, are described in detail in this
chapter.
• Do not use a STEPLIB DD statement in your production NetView job control language (JCL). A STEPLIB
statement can cause unnecessary I/O during NetView execution. See “STEPLIB DD Statements” on
page 107.
• Use the TASKUTIL command to monitor NetView task utilizations, queue lengths, storage use, and
active command lists. See “TASKUTIL Command” on page 109.
• To optimize performance for command security authorization, use AUTOSEC=BYPASS and SEC=BY on
CMDDEF statements for commands that do not require security checks. See “Command Security ” on
page 95.
• Use the OMIT operand of the STATOPT statement to control the storage requirement of the status
monitor. See “Status Monitor STATOPT Filtering” on page 107.
• Use the high performance transport instead of the management services transport for LU 6.2
communication when possible. See “LU 6.2 Transport” on page 99.
• To improve CPU usage when using installation exits, do not use dummy exits, and optimize the
performance of frequently invoked exits. See “Installation Exits” on page 98.
• Decide whether to use persistent or nonpersistent sessions for NetView-NetView communications.
Persistent sessions are preferred for all but very low traffic environments. See “Persistent and
Nonpersistent LUC Sessions” on page 103.
• You might be able to shorten the path length required to perform span of control verification by
migrating to the NetView span table. For information on implementing the NetView span table, refer to
the IBM Z NetView Security Reference.
• Use the DSRBS command to monitor the DSRB allocations for the NetView data services tasks (DSTs).
See “Data Services Request Blocks (DSRBs)” on page 97.
• To save storage, delete command definitions relating to NetView functions that you do not use. See
“Minimizing Storage Usage” on page 135.
• Do not specify the MAXSESS keyword on the CNMAUTH statement in DSILUCTD. This enables the
NetView program to allocate as many LUC sessions as it needs for alert forwarding, remote database
retrieval, and status forwarding. See “MAXSESS Keyword” on page 100.
• To reduce processor usage, consider limiting NCCF TRACE options. See “NCCF TRACE Options” on page
100.

Address Space Dispatch Priority

Because they interact closely, use the same dispatch priority for the NetView, GMFHS, RODM, and Event/
Automation Service (IHSAEVNT) address spaces.

© Copyright IBM Corp. 1997, 2022 91

Automated Operations Network (AON) Performance
Considerations
• The ENVIRON SETUP control file entry identifies attributes of NetView running AON and enables you to
tailor the control file to your needs. The ENVIRON SETUP entry is optional. If not included, AON uses all
of the defaults. ENVIRON SETUP parameters that might affect performance are:
GENALERT
GENALERT=Y is required to generate alerts. The default is N. To update the NetView management
console and Resource Object Data Manager (RODM) with automation information, specify
GENALERT=Y. If you do not require this information, a significant reduction in CPU time can result
when GENALERT=N is specified instead of GENALERT=Y.
RODMAIP
RODMAIP defines whether the RODM AIP (Automation in Progress) operator status is set, causing
the AIP pattern to display for the affected resource in the NetView management console.
RODMAIP=NO is the default setting. Specify RODMAIP=NO in the ENVIRON SETUP entry when AIP
status is not required; this will cause a reduction in the CPU time used to process the AON workload.
TRACE
TRACE=ON enables AON to perform tracing. The default value is NONE, which prevents program
entry, exit, or tracing. A setting of TRACE=OFF enables the trace facility, but no tracing is done at
this time. If you are not planning to perform tracing, setting TRACE=NONE will result in a reduction
in CPU time compared to setting TRACE=OFF.
• The ENVIRON DDF control file entry defines status update characteristics for the environment. The
ENVIRON DDF entry is required to implement DDF. The default is DDF=NO. When DDF=STATUS is
defined, DDF colors are defined by resource automation or VTAM status in DDF entries of the control
file. Specifying DDF=NO will not initialize or log events to DDF. If DDF is not being used, setting DDF=NO
will reduce the CPU time used to process the AON workload.

Browse
If you have storage limitations, control the use of the data set member browse function because it reads
the entire data set member into storage. See “Estimating Storage Usage” on page 115 for the storage
required for the data set member browse function.
Note: This consideration does not apply to browsing the network log.
Using the BROWSE command, you can browse members on a remote NetView system. When a cross-
domain browse request is processed, the RMTCMD command is used internally to satisfy the request.
The RMTMAXL parameter of the DEFAULTS and OVERRIDE commands specifies the maximum number
of lines transferred for a cross-domain member browse request. If the remote member contains more
than the maximum number of lines, the BROWSE command continues with the permitted number of lines,
and message CNM206I is issued. The BROWSE command uses the RMTMAXL setting of the operator
issuing the cross-domain browse request. A large value for RMTMAXL allows a cross-domain member
browse request to return large amounts of data, but can cause delays with other RMTCMD LU 6.2
communications. The default value of RMTMAXL is 2500 lines.

Canzlog Archiving
When you set up archiving of Canzlog data, consider the following items:
• “Canzlog Data Set Characteristics” on page 93
• “Canzlog Data Access” on page 93
• “Canzlog Archive Storage Requirements” on page 93

92 IBM Z NetView: Tuning Guide

Canzlog Data Set Characteristics
Consider the following data set characteristics when setting certain archive data set parameters in the
NetView program or the Storage Management System (SMS), particularly parameters that are related to
space and block size.
• Message data sets in a Canzlog archive have a logical record length of 1024 bytes, and contain 8192
records.
• Index data sets have a logical record length of 80 bytes and, except for the primary index data set, are
limited to 4096 records.
• The number of records in the primary index data set is not limited. Records in this data set accumulate
slowly (approximately one record every 4 weeks at a sustained rate of 50 messages per second).
Also, consider the characteristics of the devices on which the data sets are to be stored.
See the descriptions of the following statements in the IBM Z NetView Administration Reference:
• ARCHIVE.INDEX.SPACE
• ARCHIVE.INDEX.BLOCKSIZE
• ARCHIVE.MESSAGE.SPACE
• ARCHIVE.MESSAGE.BLOCKSIZE
If you use SMS to define data set characteristics for the archive data sets, see z/OS DFSMSdfp Storage
Administration.
The NetView program supplies default space and block size values that are suitable for an IBM 3390
direct access storage device. If you use a different type of device, review the device specifications,
including organization of data and device capacity; then, choose space allocations and block sizes to
ensure that complete data sets can be stored in the most efficient manner on the devices.
You can also use z/OS components such as DFSMDdfp and SMS in selecting suitable block sizes for the
data sets.

Canzlog Data Access

If many operators (50 or more) are expected to simultaneously browse or search a large amount of
archived Canzlog data, consider the following actions:
• Allocating at least the message data sets on DASD with a faster access time, especially if these
operators search the data.
• Increasing the value specified for the ARCHIVE.BROWSE.DATASPACES or
ARCHIVE.BROWSE.MAXDSPSIZE statement in the CNMSTUSR or CxxSTGEN member. For more
information about these statements, see the IBM Z NetView Administration Reference. For information
about changing CNMSTYLE statements, see IBM Z NetView Installation: Getting Started.

Canzlog Archive Storage Requirements

The amount of disk space that is required to archive Canzlog data depends on a number of factors,
including the following factors:
• The rate at which messages are produced on the system
• Storage device capacity and data organization
• The length of the time that archived data is to be available for NetView operators to browse
• The length of the time that archived data, other than the data for NetView operators to browse, is to be
kept
To estimate the required disk space, determine the rate at which messages are produced on the system.
Because message production varies during operation, examine the message production for several time
periods throughout the day or week to determine the average message production rate to use in the

Chapter 10. Additional Tuning Considerations 93

 formulas that follow. Increase or decrease that estimate based on expected future message production,
such as for increased workloads.
To illustrate the calculations for expected storage usage for an archive, consider a z/OS system on which
messages are produced at a sustained rate of 50 per second by the operating system and any jobs
running there. Also, suppose that each message line requires an average of 256 bytes of space in the
Canzlog log.
The average amount of space that is required for each message can vary due to things such as the
location of multiline messages in the log or the saving of data that occurs when a NetView program
instance that archives Canzlog data stops. However, the average should be useful for a lengthy,
continuous period of system operation with the NetView program running.
Each message data set holds an 8-megabyte (8,388,608 bytes) segment of Canzlog data, which is stored
as 8192 1024-byte records; that is, each 8-megabyte segment contains approximately 32768 messages,
calculated by using the following formula:
8,388,608 bytes per segment ÷ 256 bytes per message
At a sustained rate of 50 messages per second, each 8-megabyte segment holds messages that cover
a time period of approximately 655 seconds, or about 10.9 minutes, calculated by using the following
formula:
32,768 messages per segment ÷ 50 messages per second
At this message rate, after the NetView program begins archiving Canzlog data, an older 8-megabyte
segment of Canzlog data becomes eligible for archiving each time the latest 8-megabyte segment
becomes full, approximately every 10.9 minutes. When the latest 8-megabyte segment is full, a new
message data set is created, the data from the older 8-megabyte segment is written to the new message
data set, and a record for the new message data set is written to the current index data set. If a new
message data set is created approximately every 10.9 minutes, approximately 132 message data sets are
created per day, calculated by using the following formula:
1440 minutes per day ÷ 10.9 minutes per message data set
If you use the default space allocation and block size values that are provided by the NetView program
(11 cylinders and 27648 bytes, respectively) and if you use a 3390 device, approximately 1452 cylinders
per day are required for archiving the message data, calculated by using the following formula:
132 message data sets per day × 11 cylinders per message data set
For each message data set that is created, the NetView program adds one 80-byte record to an index data
set. An index data set, not including the primary index data set, contains a maximum of 4096 records.
At the message rate that was described previously, an index data set fills, and a new one is created,
approximately every 31 days, calculated by using the following formula:
4096 message data set records per index data set ÷ 132 message data sets per day
If you use the default space allocation and block size values that are provided by the NetView program (6
tracks and 27,920 bytes, respectively) and if you use a 3390 device, approximately 6 tracks of space per
31 days (a month) are required for a new index data set.
Each time a new index data set is created, the NetView program adds one 80-byte record for that index
data set to the primary index data set. If you use the default values for space allocation (6 tracks) and
block size (27,920 bytes) and if you use a 3390 device, two 27,920-byte blocks fit on a track. Therefore,
a 6-track extent of the primary index data set holds 4188 records, calculated by using the following
formula:
6 tracks × (2 blocks per track × (27,920 ÷ 80) records per block)
If a new index data set is created approximately every 31 days, then that 6-track extent of a primary index
data set requires approximately 129,828 days (over 355 years) to fill completely, calculated by using the
following formula:
4188 records per 6-track extent × 31 days per record for an index data set

94 IBM Z NetView: Tuning Guide

 If your storage devices for archive data sets have different capacities and data organizations than those
of 3390 devices, replace the applicable numbers in the formulas (such as blocks per track and block size)
with values that are appropriate for your storage devices. If you do not use SMS to provide the data set
characteristics for data set allocation, you might have to change one or more of the following values so
that the NetView program can allocate data sets that are more appropriate for your devices:
• ARCHIVE.INDEX.SPACE
• ARCHIVE.INDEX.BLOCKSIZE
• ARCHIVE.INDEX.UNIT
• ARCHIVE.MESSAGE.SPACE
• ARCHIVE.MESSAGE.BLOCKSIZE
• ARCHIVE.MESSAGE.UNIT
After you determine the disk space requirements for the primary index data set, the index data sets, and
the message data sets, which are based on the message production rate on your system, determine how
long the data must be kept and how much of it is to remain available for NetView operators to browse.
When archived Canzlog data is no longer required, you can delete the message data sets that contain
data that is no longer needed. Do not delete index data sets. As a result, the calculation of the disk space
requirement must include an estimate of the total number of index data sets that might be created over
an estimated useful life of the system and the NetView product.
For this example, to complete the calculations, consider an estimated useful life of approximately 20
years (7305 days). If you use 3390 devices for archived Canzlog data, and if the sustained message
production rate on this system is 50 messages per second, and if you keep 2 months of message data
(deleting message data monthly after 2 months of data are collected, which means that nearly 3 months
of message data is available just before a deletion occurs), the following storage is required:
• For the index components: 95 cylinders, calculated by using the following formula:
(6 tracks for a primary index data set + 1416 tracks for all other index data sets) ÷ 15 tracks per
cylinder
The tracks that are needed for the index data sets are as follows:
– 1 primary index data set: 6 tracks is sufficient
– 236 index data sets: 1416 tracks
236 index data sets is calculated by using the following formula:
7305 days of useful life ÷ 31 days per index data set
1416 tracks is calculated by using the following formula:
236 index data sets × 6 tracks per data set
• For the message components (message data sets): 135,036 cylinders, calculated by using the following
formula:
1452 cylinders per day × 31 days per month × 3 months
• For the whole archive: 135,131 cylinders, calculated by using the following formula:
95 cylinders for index components + 135,036 cylinders for message components
Note that many of the numbers in this section are estimates. You might want to increase some of the
percentages.

Command Security
This section discusses certain considerations for achieving maximum performance of NetView when
security checks are run against commands. There are two different methods of command security in
NetView: the NetView command authorization table and a system authorization facility (SAF) product,
such as RACF®. See the IBM Z NetView Security Reference for more information about command security.

Chapter 10. Additional Tuning Considerations 95

 Considerations:
• Regardless of which command security you use, you can code SEC=BY on a CMDDEF statement in
the CNMCMDU initialization member for commands that do not require a security check (HELP, for
example). This improves performance by eliminating security check processing time on commands that
are safe to use in your environment.
• For all three methods of security, you can code AUTOSEC=BYPASS on the DEFAULTS command
to bypass security checks on all commands originating from the automation table, assuming your
automation table member update process is secure. This can eliminate unnecessary checking of
commands not entered at an operator console.
• When using RACF, auditing all resources can degrade system performance. Setting RACF auditing to
NONE for resources in the NETCMDS class can improve performance. Using RACF AUDIT provides you
with an audit trail of attempts to issue unauthorized commands or command lists.
• When using a backup command authorization table with SAF as your primary command security
method, you should make sure generic command identifiers are specified to prevent unnecessary
searches of the backup command authorization table. For example, in RACF:

RDEFINE NETCMDS *.*.* UACC(READ)

• When migrating to the SAF NETCMDS class or the NetView command authorization table with
the SECMIGR command, excess statements might be generated which can be deleted to (slightly)
improve performance. When the SECMIGR tool generates statements equivalent to scope KEYCLASS
statements, and a VALCLASS statement is not specified for the keyword, a statement is generated to
cover any specified values. For keywords that cannot have values, this statement can be safely deleted.
For example, because the AUTOTBL OFF keyword has no value, statements generated for a command
identifier of netid.luname.AUTOTBL.OFF.* can be safely deleted.
• Processing generic command identifiers is the most performance intensive part of searching the
command authorization table. If you are using generic command identifiers (wildcards) in the command
authorization table, you can code specific command identifiers on EXEMPT statements for commands
that do not need protection. Using specific (not generic) PROTECT statements for frequently used
commands should also be helpful.

Command Statistics

Enabling the Command Statistics function introduces some overhead. To minimize the overhead, consider
the following items:
• Run command statistics with the PRIONLY option.
• Limit the commands that are monitored by configuring an inclusion and/or exclusion list in the
Command Statistics Inclusion/Exclusion (CNMSCSIE) sample.
• Dynamically enable and disable the command statistics function with the CMDMON STATS=ALL/
PRIONLY and CMDMON STATS=OFF commands if you are using the function only for problem diagnosis
for one or more commands.
• Specify a minimum value of 10000 for the CMDMON.DATA.MAXRECS statement in your CNMSTUSR or
CxxSTGEN member.
– If you have storage limitations, each command statistics record uses 108 bytes of storage.
• If you don’t need to retrieve Command Statistics data from SMF logs, set LOGSMF=NO for the CMDMON
command, or specify CMDMON.INIT.LOGSMF=NO statement in CNMSTYLE to disable SMF logging.
• Specify a shorter interval in the SMF SMFPRMxx parmlib member if you need to retrieve Command
Statistics data from SMF logs, which can improve the storage efficiency when writing SMF records.

96 IBM Z NetView: Tuning Guide

Data Services Request Blocks (DSRBs)
The following types of data services request blocks (DSRBs) are used to store information about a
transaction request being processed by a data services task (DST):
• Unsolicited (DSRBUs)
• Solicited (DSRBOs)
DSRBUs are used to process unsolicited problem determination request units (RUs). DSRBOs are used to
process solicited RUs and VSAM requests. If more requests are received than there are DSRBs available,
the requests are queued.
The number of DSRBOs and DSRBUs allocated for a DST are defined as parameters on DSTINIT
statements in the DSIPARM initialization member for the DST. If you do not specify the parameters,
the default allocations are 3 DSRBOs and 5 DSRBUs. See the IBM Z NetView Administration Reference for
the syntax of DSTINIT statements.
The DSRBS command displays statistics on the data services request block use for the NetView program
and user-written DSTs. The DSRBS output is a snapshot of the current DSRB use, and can be used to
determine if the DSRB allocations are sufficient.
Figure 21 on page 97 shows a sample of output from the DSRBS command.

DSRBS Data Services Request Block Usage for BNJDSERV 13:35:07

Unsolicited DSRBs: 5 Used: 0 Free: 5

Solicited DSRBs: 5 Used: 5 VSAM Redrive: 0 Free: 0
TOTAL DSRBs: 10 Used: 5 VSAM Redrive: 0 Free: 5

Current DSRB Usage

1 2 3 4 5 6 7 8
No. DSRB STATUS Taskname Type Request Redrive Serial No. Step No.
001 Unsl Inact
002 Unsl Inact
003 Unsl Inact
004 Unsl Inact
005 Unsl Inact
006 Soli Active NCF01PPT VSAM Erase No 5104 5
007 Soli Active NCF01PPT VSAM Get No 5100 14
008 Soli Active NCF01PPT VSAM Put Yes 5105 8
009 Soli Active PCF01PPT VSAM Get No 5102 13
010 Soil Active PCF01PPT VSAM Put No 5103 12

Figure 21. Sample Output from the DSRBS Command

The statistics displayed include:

DSRB Type
Statistics
Unsolicited DSRBs (DSRBU)
Number allocated, in use, and available
Solicited DSRBs (DSRBO)
Number allocated, in use, being redriven because of VSAM contention, and available
Total DSRBs (DSRBO + DSRBU)
Number allocated, in use, being redriven because of VSAM contention, and available
For each DSRB currently in use, the following information is displayed:
1
Type of DSRB (unsolicited or solicited)
2
Status (active or inactive)
3
Task that initiated the request

Chapter 10. Additional Tuning Considerations 97

 4
Type of request (VSAM or CNM)
5
Type of VSAM request (ENDREQ, ERASE, GET, POINT, PUT)
6
Whether the VSAM request is being redriven because of contention (YES or NO)
7
Request serial number
8
Step sequence number
The DSRBS command is useful in determining DSTs that are waiting for requests to complete. DSRBS is
also useful in determining the optimum number of DSRBs to allocate. Too few DSRBs can result in DST
requests being queued. Too many DSRBs can result in an excessive numbers of VSAM requests being
redriven, which causes extra CPU overhead.
Consider the following items for tuning the DSRB allocations for a DST.
• If DSRBOs are frequently in VSAM redrive status, consider lowering the number of DSRBOs allocated for
the DST.
• If a majority of the DSRBs are frequently in use, consider increasing the number of DSRBs allocated for
the DST.
• Monitor the DSRB usage for the DSIGDS, AAUTSKLP, and AAUTCNMI tasks closely, because they benefit
from additional DSRBs in many environments.
• Performance of the BNJDSERV task (hardware monitor) can degrade if additional DSRBs are added,
because its VSAM requests might go into redrive status.

Installation Exits
This topic provides performance considerations for using exits.
• For most OST-type exits (DSIEXxx), a LOAD FAILED message is issued if the exit is not found. Using
dummy exits would prevent this message, but degrade system performance by causing the NetView
program to process instructions to set up a call to the dummy exit every time the exit is driven, therefore
unnecessarily using CPU.
• Coding exits for frequently called exits also degrades performance on systems with heavy message
loads, especially if the exits are coded in C or PL/I. C has a greater initialization overhead than PL/I. If
you code very frequently called exits, consider using Assembler.
• If you use the BLOG command list, be aware that DSIEX18 is run under the browse task to match the
search arguments specified in the BLOG input. If you have a large log and the search string matches
only a small number of records, the forward or backward function key causes the log to be searched
until enough records matching the string are found to fill the screen. This could mean that thousands of
records will be searched before enough matches are found to fill the screen, while resulting in a large
amount of processing time being used.

LOGTSTAT Command
The LOGTSTAT command can be used to write task utilization data to the System Management Facility
(SMF) log. You can use the LOGTSTAT command to create a record for one specified task, or for all tasks
that were running at the time when the LOGTSTAT command was issued. If LOGTSTAT is used to generate
records for all tasks, an SMF record is written for each task that is active.
LOGTSTAT data can be useful in determining task start and end times. It also provides some Resource
Limits data similar to that produced using TASKMON. Specific system storage usage for a task from
startup to termination or for a known time interval can be determined from the LOGTSTAT data as well.
For more information, see the IBM Z NetView Command Reference Volume 1 (A-N) or the NetView online
help.

98 IBM Z NetView: Tuning Guide

LU 6.2 Transport
The NetView LU 6.2 transport is a programming interface that implements architected protocols to enable
applications in network nodes to communicate using conversations over LU 6.2 sessions.
The NetView LU 6.2 transport consists of two similar application program interfaces: the management
services (MS) transport and the high performance transport. For applications that run in the NetView
program, each transport provides a high-level programming interface to mask the LU 6.2 complexities. An
application registered with the appropriate transport can send data in architected envelopes to a partner
application and receive data in return.
Although both transports provide the same functions and mask the LU 6.2 complexities, each transport
offers its own advantages.
The high performance transport uses different LU 6.2 protocols that are faster than the protocols used
by the MS transport. Because of these protocols, the high performance transport provides general error
notification rather than specific error notification about data. In addition to the advantage of speed, the
high performance transport enables programmers to define session parameters such as RU size.
The MS transport uses LU 6.2 conversation protocols that generate more network traffic than the high
performance transport protocols to transport each piece of data. The advantage of the MS transport is
that it guarantees delivery of data or specific error notification about the data.
See the IBM Z NetView Application Programmer's Guide for information about LU 6.2 transports.
The following tuning considerations apply to the LU 6.2 transport.
• When designing applications to use the LU 6.2 transport, use the high performance API instead of the
management services API when possible.
• Send requests with a reply expected require more processing than send requests without reply
expected, because of the extra overhead of timeout checking and processing the reply. When designing
applications to use the LU 6.2 transport, use send requests without reply expected where possible.
• To use send requests with reply expected, create a CMDDEF statement for the reply processor in the
CNMCMDU NetView initialization member. If the reply processor is a command list, use the following
statement:

CMDDEF.clistnameL.MOD=DSICCP

The presence of the CMDDEF statement eliminates I/O to the command list data set (DSICLD) to verify
the existence of the reply processor before the request is sent.
• The NetView constants module (DSICTMOD) contains an entry for the LU 6.2 transport support. This
entry specifies the number of LUs with which you expect to have sessions. This value is used to optimize
control block access. The default value is 2000. If you expect to have more partner LUs than the default,
change this value in DSICTMOD. The number you specify does not need to be exact, but too small a
number hinders access to control blocks. A value in excess of the expected number of partner LUs can
result in unused virtual storage, but can improve access to control blocks. In general, it is better to
overestimate than underestimate. See the IBM Z NetView Installation: Getting Started for information
about changing and relinking DSICTMOD.
• The high performance transport enables applications to specify their logmode when they register.
Sample logmode definitions are contained in member CNMS0001. Logmodes have an RUSIZES
parameter that you can use to specify the maximum size of data in bytes that the LUs can send. The
default RUSIZES parameter for LU 6.2 applications is 8585. The first two numbers are for the primary
LU, and the second two numbers are for the secondary LU. Each pair of numbers represents a mantissa
and an exponent, as follows:

M × 2N

The default 8585 means that both the primary and secondary LU can send a maximum of 8 x 25, or
256 bytes. Adjust the RUSIZES parameter appropriately for your LU 6.2 applications. The RMTCMD
command uses the PARALLEL logmode (in CNMS0001), which uses RUSIZES=8787 (or 1024 bytes).

Chapter 10. Additional Tuning Considerations 99

 Consider increasing the RUSIZES parameter for the PARALLEL logmode to higher values, so that the
RMTCMD can send larger data buffers.
• To forward messages or management service units (MSUs) to the generic automation receiver (NVAUTO)
for automation table processing, modify the command definition for DSINVGRP to make the generic
automation receiver command processor resident. To do this, add the following to the CNMCMDU
initialization member:

CMDDEF.DSINVGRP.RES=Y

Making this command processor resident avoids the I/O needed to load the command processor every
time the generic automation receiver processes a message or MSU.

MAXSESS Keyword
You can use the MAXSESS keyword on the CNMAUTH statement in DSILUCTD to restrict the number of
cross-domain sessions the NetView program can set up to an adjacent domain. The value for MAXSESS, if
specified, can be in the range 1-65535. The value in the samples is 10.
If you do not specify the MAXSESS keyword, the NetView program allocates as many sessions as it needs
for alert forwarding and remote database retrieval for the hardware monitor and session monitor. Do
not specify a value for MAXSESS unless you need to restrict the number of cross-domain sessions the
NetView program can set up.
If you use nonpersistent sessions, do not specify a value for MAXSESS, because any idle sessions are
brought down.
If you must specify a value for MAXSESS, use the following formula to calculate a value:

MAXSESS value = 2 (Number of session monitor sessions)

+ 1 (Alert forwarding, if NPDA.ALERTFWD=NV-UNIQ in CNMSTYLE

%INCLUDE member CNMSTUSR)

+ xx (Number of concurrent NetView operators doing

hardware monitor remote data retrieval)

If the MAXSESS value is exceeded, an SDOMAIN (set domain) command from the session monitor and
hardware monitor can fail and alerts might not be forwarded.
If an SDOMAIN command fails, message DSI784 is issued.

NCCF TRACE Options

To help reduce processor consumption, especially for autotasks, consider running without the STOR
option on the NCCF TRACE command. If the OPTION keyword is not specified on the TRACE ON keyword,
the default options of QUE, PSS, DISP, STOR, and UEXIT are used. These are required for debugging
by IBM Software Support. However, if your environment is heavily automated with autotasks, consider
specifying the TASK= option on the TRACE command and specifying all task types except OST. Tracing is
then inactive for all OSTs and AOSTs (autotasks). The processor usage is reduced because of the STOR
option and associated processor consumption for autotasks. However, pertinent problem determination
and analysis data can be lost. Users might be requested by IBM Software Support to temporarily
reactivate tracing for OSTs and recreate with TRACE active for all task types. For additional information
regarding the NCCF TRACE command, see the IBM Z NetView Command Reference Volume 1 (A-N) or the
NetView online help.

NetView Access from the Web Browser

This function provides the capability for NetView users to convert host data to HTML and then make
it available to web browsers by way of the IBM Internet Connection Server. Users can log on to the
internet from a workstation, go to a URL for their NetView system, and request/receive command-line
data responses.

100 IBM Z NetView: Tuning Guide

 There is a 1000 line limit for allowable data display when using the web browser function. If more than
1000 lines of data are requested, users see a message stating how many lines were truncated. Users
should not knowingly attempt to browse members or data sets that far exceed this limit. The entire
file must be read in, causing an increase in response time and CPU usage, both at the host NetView
program and at the workstation. Performance testing showed no significant difference in response time to
commands with regard to both platform and specific web browser used.

NetView Constants Module (DSICTMOD)

DSICTMOD is the NetView constants module. The sample is CNMS0055. This sample assembles and
link-edits the module. Changes in the module require an assembly and link-edit.
All of the values are described in IBM Z NetView Installation: Getting Started. The following constants are
described in this book:
• The expected number of task global variables is described in “Global Variables” on page 34.
• The expected number of common global variables is described in “Global Variables” on page 34.
• Sense code filtering is described in “DASD Filtering” on page 61.
• LU 6.2 transport values are described in “LU 6.2 Transport” on page 99.
• Nonpersistent session timeout values are described in “Persistent and Nonpersistent LUC Sessions” on
page 103.
• The management of below-the-line storage is described in “Minimizing Storage Usage” on page 135.

NetView-NetView Communication
You can use the RMTCMD command to send commands to another NetView program and receive
responses, and to query details about RMTCMD associations. The RMTCMD command offers the following
advantages over NetView-NetView (NNT) sessions and the ROUTE command.
• The RMTCMD command consolidates the LU-LU communications from multiple operators by sending
commands to a remote NetView program using a pair of LU 6.2 sessions. This saves storage and
processing time for setting up sessions, and for maintaining session awareness in VTAM and the session
monitor. A separate session is not needed for each operator.
• The RMTCMD command uses large buffers and RU sizes, improving performance for operators receiving
large multiline messages.

NetView Program-to-Program Interface

The NetView program-to-program interface enables application programs to send network management
vector transport (NMVT) or control point management services unit (CP-MSU) requests to the NetView
program and enables application programs to send data buffers to, or receive data buffers from, other
application programs. An application program can be a sender program, a receiver program, or both.
Application programs can send data buffers to, or receive data buffers from, other applications running in
the same host as the NetView program, or in a different host.
Each receiver program has a buffer queue for temporary storage of incoming data buffers. These buffer
queues reside in the program-to-program interface. A sender program sends a data buffer to a receiver
buffer queue, and the receiver program retrieves the data buffer from the buffer queue.
When you define a program as a receiver, you also define the buffer queue limit, the maximum number of
outstanding buffers that can be stored in the receiver buffer queue. When the receiver buffer queue is full,
and other buffers are sent, the sender programs receive a return code of 35.
You can use the DISPPI command to display information about PPI buffer queues, including buffer limits,
buffer queue lengths, total buffers sent, and buffer storage usage. This command displays information
for the current receiver or receivers defined to the program-to-program interface. Figure 22 on page 102
contains sample output from the DISPPI command.

Chapter 10. Additional Tuning Considerations 101

 DWO948I RECEIVER RECEIVER BUFFER QUEUED TOTAL STORAGE
DWO949I IDENTITY STATUS LIMIT BUFFERS BUFFERS ALLOCATED
DWO950I -------- -------- ---------- ---------- ---------- ----------
DWO951I NETVALRT ACTIVE 1000 0 18 0
DWO951I DUIFSSCO ACTIVE 100 0 10049 0
DWO951I PPICMDID ACTIVE 100 0 0 0
DWO951I DUIATMGR ACTIVE 1000 0 3287 0
DWO951I FNMRCVQ ACTIVE 99 0 20111 0
DWO951I NPM60AP ACTIVE 99 0 20111 0
DWO951I DUIFI000 ACTIVE 1000 0 15027 0
DWO968I END OF DISPLAY

Figure 22. Sample DISPPI Command Output

Another command, DISBQL, displays buffer queue limits and the number of buffers currently available on
the buffer queue. You can use the SETBQL command to reset the buffer queue limit of a receiver.
The program-to-program interface trace facility enables you to set up a trace in the program-to-program
interface for either an individual receiver or all current and future receivers. The program-to-program
interface trace facility writes a trace record each time a user defines, deactivates, or deletes a receiver
to the program-to-program interface and each time a buffer is sent or received. You control the program-
to-program interface trace facility with the TRACEPPI command. You can use the TRACEPPI command
to start, stop, modify, or end the program-to-program interface trace facility. See the IBM Z NetView
Command Reference Volume 1 (A-N) or the NetView online help for more information about the TRACEPPI
command.
See the IBM Z NetView Application Programmer's Guide for detailed information about the NetView
program-to-program interface.

Network Asset Management Facility

The network asset management facility enables you to collect vital product data (VPD) from the
network. See the IBM Z NetView Command Reference Volume 1 (A-N) or the NetView online help for
information about these commands: VPDALL, VPDPU, VPDDCE, VPDCMD, and VPDLOG. The nature of
those commands and the functions they perform extend their elapsed time for execution.
A request for VPD is sent serially to the physical units (PUs) and data circuit-terminating equipment (DCE)
in the network. The PUs and DCEs must receive the request and send a response. The elapsed time for
this collection varies depending on the configuration of the network and the traffic rate.
Consider dividing your network into logical sections and scheduling the collection of VPD from a section at
a time during off-peak hours. See the IBM Z NetView Application Programmer's Guide for interpretation of
messages returned from the VPDCMD command.

Partitioned Data Set (PDS) Allocation

The following are the non-VSAM data sets used by the NetView program.
BNJPNL1
NetView hardware monitor panel data sets
BNJPNL2
NetView hardware monitor panel data sets
CNMPNL1
NetView panel data sets
DSICLD
NetView command list data sets
DSILIST
NetView listing data sets
DSIMSG
NetView messages data sets

102 IBM Z NetView: Tuning Guide

 DSIPARM
NetView definitions, automation table, and CMDDEF statements
DSIPRF
NetView operator profile definitions
DSIVTAM
VTAMLST data sets

The command list data set (DSICLD) is the most critical data set from a performance standpoint.
Each time a command list is selected for execution, it is located and read from that data set (unless
it is preloaded by LOADCL). A normal BLDL search of the partitioned data set directory is done to
find the command list, enabling you to make command list changes dynamically. Usually, DSICLD is a
concatenation of several data sets that can be on different direct access storage device (DASD) volumes.
Define the command list data sets with one extent and ensure that the extent will not fill up over the time
the NetView program is running. If a data set expands to multiple extents after the NetView program is up
and running, attempts to access members newly placed in the additional extents can cause I/O errors.
Retrieval time for command lists is an important consideration in command list performance. You can
avoid physical I/O either by preloading command lists using the LOADCL command or by using a virtual
I/O data set for DSICLD. If you cannot use either of those options, you can minimize physical I/O response
time by placing DSICLD on a cached DASD device. If some of your command lists are on cached DASD and
you have multiple volumes, put the cached volumes first in the DD definition list for DSICLD.
You can create temporary VIO data sets by placing an IEBCOPY statement in the NetView startup
procedure, which copies the command lists to a VIO data set and passes that data set on to the NetView
program. When you use these temporary data sets, you cannot dynamically change the command lists.
You must restart the NetView program to include the changes.
If you use panels extensively, you can reduce I/O delays by placing the panel data sets in a VIO data set.
However, you cannot change such panels dynamically.

Persistent and Nonpersistent LUC Sessions

If you have a multiple-domain environment with copies of the NetView program on different hosts
communicating with one another, you need to decide whether to use persistent or nonpersistent sessions
for the NetView-NetView communication used by the session monitor, the hardware monitor.
Persistent sessions are most commonly used with leased lines, while nonpersistent sessions are most
commonly used with dialed (switched) lines. Using nonpersistent sessions involves the additional
overhead of establishing the dialed connection. Therefore, use nonpersistent sessions only when you
expect very low traffic over the session. In deciding between persistent and nonpersistent sessions,
examine the trade-off between leased line cost, the host processor cost of establishing the dialed
connection, and the nonpersistent session.

Using Nonpersistent Sessions over Dialed Lines

If you decide to use nonpersistent sessions over dialed lines, you should understand the relationship
between the timers maintained by the NetView program and NCP.
• For every nonpersistent session that it maintains, the NetView program has an activity timer, which it
resets every time traffic is sent over the session. The value for this timer is defined by a constant in
DSICTMOD. See IBM Z NetView Installation: Getting Started for more information. If there is no activity
over the session within the timer interval, the NetView program ends the nonpersistent session.
• For dialed connections involving an NCP (VTAM-NCP or NCP-NCP), NCP has an activity timer, which
it resets every time traffic is sent over the dialed line. This NCP timer is controlled by setting
BRKCON=CONNECTO on a line or PU statement in the NCP definition. The value for this timer is set by
the ACTIVTO parameter on the GROUP statement. Refer to the appropriate NCP, SSP, and EP publication
for more information about activity timers. If no traffic occurs over the dialed line within the timer
interval, NCP ends the dialed connection.

Chapter 10. Additional Tuning Considerations 103

 • For VTAM-VTAM dialed connections, if VTAM detects that a dialed connection is active but no sessions
are occurring over it, VTAM ends the dialed connection.
For a nonpersistent session over a VTAM-VTAM dialed connection, the session times out according to the
NetView timer. The amount of time before the line drops is therefore determined by the NetView timer.
If NCPs are involved in the dialed connection, the nonpersistent session times out as follows:
• If the NCP timer value is less than the NetView timer value, the NCP ends the dialed connection, causing
the nonpersistent session to end. In this case, the line drop time is based on the NCP timer.
• If the NetView timer value is less than the NCP timer value, the NetView program brings down the
nonpersistent session. Because the session termination causes traffic over the dialed connection, NCP
resets its activity timer. If no other activity occurs within the NCP timer interval, NCP ends the dialed
connection. In this case, the line drop time is based on the NetView timer plus the NCP timer.

RESOURCE Command
The RESOURCE command displays statistics about NetView system resource use. This information is
helpful in determining the amount of system resources used by the NetView program. Figure 23 on page
104 contains sample output from the RESOURCE command.

DSI386I NETVIEW RESOURCE UTILIZATION 09:58:07

1 TOTAL CPU % = 6.65
2 E530EENV CPU % = 0.81
3 E530EENV CPU TIME USED = 14.06 SEC.
4 REAL STORAGE IN USE = 7664K
5 PRIVATE ALLOCATED < 16M = 612K
6 PRIVATE ALLOCATED > 16M = 17596K
7 PRIVATE REGION < 16M = 4160K
8 PRIVATE REGION > 16M = 32768K
END OF DISPLAY

Figure 23. Sample Output from the RESOURCE Command

The following information is displayed in the command output (see Figure 23 on page 104).
1
Total CPU Utilization
Total complex CPU utilization based on a maximum of 100%. This utilization is calculated over the
most recent 1 second interval.
2
NetView CPU Utilization
NetView CPU utilization based on a maximum of 100%. This utilization is calculated over the most
recent 1 second interval.
3
NetView CPU Time Used
The combination of task control block (TCB) and service request block (SRB) CPU time used. This field
is cumulative from when the NetView program was first started.
4
Real Storage In Use
The number of real storage frames that are in use (shown in kilobytes).
This value includes the number of real storage frames that are in use for both the storage that is
allocated in the NetView address space and the data spaces that are owned by the NetView program.
5
Private Allocated Below 16 M
The amount of virtual storage allocated below the 16 MB line.

104 IBM Z NetView: Tuning Guide

 6
Private Allocated Above 16 M
The amount of virtual storage allocated above the 16 MB line.
7
Private Region Below 16 M
The total amount of virtual storage below the 16 MB.
8
Private Region Above 16 M
The total amount of virtual storage above the 16 MB.
The CPU use of the NetView program depends on message and network traffic levels, which can appear
in bursts. Therefore, the CPU utilization can appear to spike over the small 1 second interval that the
RESOURCE command uses. The cumulative CPU time, monitored over time, is a better indicator of the
level of CPU use.
The real (processor) storage use of the NetView program includes central plus expanded storage frames
allocated to the NetView program. On systems with little or no storage contention, a real storage use can
be inflated and might not represent the true working set size.

Resource Limits
Use the NetView resource limits function to monitor and limit resource usage for various NetView tasks.
You can obtain information that you can use to plan for and tune the NetView program. You can then
adjust keywords according to the amount of storage consumed, CPU used, and other factors.

Keywords for Resource Limits

The following keywords can be adjusted to limit resource usage:
MAXSTG (Maximum Storage for a task)
Specifies the maximum amount of storage in Kilobytes that a task can use. When MAXSTG is reached
and further storage requests are made, the DSIGET macro returns an out of storage return code.
SLWSTG (When to start limiting a task from getting storage)
Specifies the maximum amount of storage in kilobytes that a task can use before slowdown measures
are used. When a task exceeds SLWSTG, all storage requests (using DSIGET) have a time delay.
AVLMAX
Specifies a percentage that determines, for any task, at which value the DSIGET macro returns an out
of storage return code. In addition, queuing a message to a task that is over its AVLMAX limit results in
the task not active condition, and the message is not transferred.
The percentage is specified as a decimal number in the range of 0–99. The NetView program
computes the ratio of the amount of storage a task is using compared to the sum of that amount
and the amount of storage left in the NetView address space. If the task usage is above the specified
limit, the DSIGET rejects the request.
Note: The specified limit enables users to have the tasks that are under the most storage stress to
fail instead of letting the entire region deplete storage, which can result in the failing of all tasks. For
example, when the default of AVLMAX=90 is used, a task using 45 MB of storage has a storage failure
if the amount of free space falls below 5 MB. Specifying a lower value for AVLMAX leaves more of
the free address space for tasks that do not overuse storage. Specifying a higher or no limit value for
AVLMAX increases the risk for storage failures on all tasks, regardless of whether they are in storage
stress.
When the NetView program is started, the AVLMAX default is set to 90. This setting can be modified
by using the DEFAULTS command.

Chapter 10. Additional Tuning Considerations 105

 AVLSLOW
Specifies a percentage that determines, for any task, at which point slowdown measures are used. In
addition, queueing a message to a task that is over its AVLSLOW limit results in that task having the
same slowdown measures applied based on how much over the limit the receiving task gets.
When the NetView program is started, the AVLSLOW default is set to 85. This setting can be modified
by using the DEFAULTS command.
Slowdown measures are used when a task exceeds AVLSLOW. All storage requests (using DSIGET)
and message queuing to the affected task have a time delay. The time delay is one microsecond for
each byte of storage requested over the SLOWSTG limit value, and quadrupled for each 1% the task is
beyond the AVLSLOW value thereafter. As the task's use of storage grows, this produces a slowdown
effect that is proportional to the size of the request.
The initial slowdown rate is calculated to allow storage to grow at a rate of 1 megabyte per second,
in the range 0–1% beyond the limit value. When the AVLSLOW value is lowered, the point at which
slowdown occurs is lowered. For example, SLOWSTG=85 triggers if the task is using 42.5 MB of
storage and only 7.5 MB of good memory is left in the region. SLOWSTG=80 triggers if the task is using
40 MB of storage and only 10 MB are left in the region.
MAXCPU (Maximum CPU for a task)
Specifies the maximum CPU utilization allowed for a task. When a task exceeds the limit, automatic
task slowdown measures are used to bring the task back into the specified range. The task is
suspended until enough time passes for the CPU to be below the specified limit.
MAXIO (Maximum input/output transactions for a task)
Specifies the maximum number of logical VSAM I/O requests per minute allowed for a task. When a
task exceeds the limit, automatic task slowdown measures are used to bring the task back into the
specified range.
MAXMQIN (Maximum number of messages received by a task)
Specifies the number of message KB per minute that is allowed to be sent to the task from other
tasks. When a task exceeds the limit, automatic task slowdown measures are used to bring the task
back into the specified range. The intent is to add enough of a delay to each request so that the
measured rate over a one minute or longer period of time falls to near or below the specified value.
MAXMQOUT (Maximum number of messages a task can send to a task)
Specifies the number of message Kilobytes per minute allowed for a task to send to another task.
When a task exceeds the limit, automatic task slowdown measures are used to bring the task back
into the specified range. If the task attempts to queue a message to another task, it is slowed down
until the rate is under the limit. The intent is to add enough of a delay to each request so that the
measured rate over a minute or longer period falls to near or below the specified value.
Changes can be made dynamically to all resource limits settings. Old values are cancelled within one
second of the OVERRIDE command being processed. The tasks continue with the new limit in force.
Coding a value of 0 turns off a specified resource limit setting. See the IBM Z NetView Command
Reference Volume 1 (A-N) or the NetView online help for more information.
LOGTSTAT (Whether to log the data to SMF)
Specifies whether resource utilization data is logged to the external log (SMF).

Using Resource Limits

Be careful when you restrict a task with resource limits. Gather performance data using the V621PERF
command list sample in Figure 1 on page 4 and then analyzed. Gather data over time for peak, standard,
and off-shift workload to ensure that a complete analysis can be made.
There are certain areas within the NetView program that use a large amount of storage, high CPU during
certain operations or at initialization. Do no limit these tasks beyond the default settings. For example,
the Session Monitor main task, AAUTSKLP, can use high CPU during purge processing and can use a large
amount of storage during a network outage. Do not limit resource utilization for tasks, such as these, at
peak workload.

106 IBM Z NetView: Tuning Guide

 The resource limit function can prevent an operator task or autotask from looping and saturating a
processor. When the task CPU limit is exceeded, BNH161I is issued and the CPU usage, for the OST, will
be throttled. User intervention can terminate the application. The same result is true for a user-written
application that is issuing DSIGET macros to obtain storage, but never using DSIFRE to release the
storage. Resource limits provide the ability to put a cap on the task storage usage and will abend the task
if the storage limit exceeds the maximum allowed (MAXSTG).

Status Monitor STATOPT Filtering

The status monitor component of the NetView program collects and summarizes information on the
status of resources defined in a VTAM domain. The status monitor can handle a maximum of 999999
resources. The status monitor preprocessor uses the resource definitions in the VTAMLST data sets to
create a DSIPARM member named DSINDEF. DSINDEF contains the network resource information that
the status monitor reads during initialization. See IBM Z NetView Installation: Configuring Additional
Components for information on the preprocessor operation.
The status monitor requires approximately 120 bytes of storage for each resource that it monitors. By
default, the status monitor keeps status for every resource in the network. If you would like to reduce the
storage requirement for the status monitor, you can use the OMIT operand of the STATOPT statement. By
specifying OMIT in the VTAMLST data sets, you can omit a node, plus all of the dependent lower nodes
that follow, from the status monitor's view of the network definition.
See the IBM Z NetView Administration Reference for more information about coding the STATOPT
statement.

STEPLIB DD Statements
Do not use a STEPLIB DD statement in your production NetView job control language (JCL). The presence
of the STEPLIB DD statement causes its directory to be searched for each LOAD, LINK, or XCTL system
macro executed during normal operation. That directory search degrades the performance of command
procedures.
Place the NetView load libraries on the system's LINKLST. See IBM Z NetView Installation: Getting Started.
For considerations on placing the HLL runtime libraries, see “Command Processors” on page 31.

TASKMON Command
The TASKMON command is a REXX procedure that provides color-coded monitoring of all NetView tasks.
The output under each group is sorted by the severity index. The Severity Index represents a percentage
of the maximum value allowed. Usage of TASKMON is similar to TASKUTIL usage. You can use the data
shown in the TASKMON output to view current and historical data on Resource Limits.
Color codes used by the TASKMON command:
White
SLOWSTG limit exceeded
Yellow
70% of limit for this line exceeded
Pink
80% of limit for this line exceeded
Red
90% of limit for this line exceeded
TASKMON provides statistics based on CPU percentage used on a single processor. TASKUTIL provides
statistics based on CPU percentage for the sum of all defined processors, for example, a sysplex with six
CPUs dedicated to the NetView program.
To issue a TASKMON command, enter:

Chapter 10. Additional Tuning Considerations 107

 TASKMON * *

A response that is similar to the following response is displayed:

TASKMON ---- START OF REPORT ----

Severity Index OPID Current Session Maximum Limit
------CPU------ -------- --------- --------- --------- ---------
152.00% OPER3 1.52 % 0.19 % 33.85 % 1.00 %
23.34% AUTO2 11.67 % 1.53 % 67.24 % 50.00 %
0.03% NTV98PPT 0.03 % 0.04 % 4.16 % 99.99 %
0.02% DSIMONIT 0.02 % 0.01 % 0.04 % 99.99 %

Severity Index OPID Current Session Seconds

-Penalty Time-- -------- --------- --------- --------- ---------
81.89% OPER3 81.89 % 2.97 % 81.66 S

Memory= 46.26% 24-Bit= 16.63% 31-Bit= 50.02% Left= 19844 K

Severity Index OPID Current Maximum Limit Slowdown

----Storage---- -------- --------- --------- --------- ---------
7.15% AAUTSKLP 1528 K 1544 K 999999 K 999999 K
5.86% AUTO2 1172 K 3605 K 20000 K 4000 K
4.93% MAINTASK 1028 K 1028 K 26214 K 23592 K
1.96% CNMTAMEL 397 K 401 K 999999 K 999999 K
1.21% AAUTCNMI 244 K 295 K 999999 K 999999 K
0.82% NTV98PPT 164 K 265 K 999999 K 999999 K
0.81% AUTO1 162 K 251 K 999999 K 999999 K
0.81% DSISVRT 162 K 213 K 999999 K 999999 K
0.47% DSIAL2WS 93 K 121 K 999999 K 999999 K
0.42% BNJDSERV 84 K 85 K 999999 K 999999 K
0.31% NTV98VMT 62 K 81 K 999999 K 999999 K
0.30% DSILCOPR 59 K 114 K 999999 K 999999 K
0.29% OPER3 57 K 60 K 999999 K 999999 K
0.28% ALIASAPL 55 K 102 K 999999 K 999999 K
0.27% DSI6DST 54 K 112 K 999999 K 999999 K
0.23% DSIGDS 45 K 89 K 999999 K 999999 K
0.22% VPDTASK 44 K 75 K 999999 K 999999 K
0.21% NTV98LUC 42 K 93 K 999999 K 999999 K
0.21% DSIHPDST 42 K 93 K 999999 K 999999 K
0.20% DSIQTSK 39 K 94 K 999999 K 999999 K
0.18% DSIQRV4A 36 K 36 K 999999 K 999999 K
0.18% DSIQRV4B 36 K 36 K 999999 K 999999 K
0.18% DSIQRV4C 36 K 36 K 999999 K 999999 K
0.17% DSIUDST 34 K 85 K 999999 K 999999 K
0.17% DSICRTR 33 K 85 K 999999 K 999999 K
0.15% DSIAMLUT 30 K 73 K 999999 K 999999 K
0.14% BNJMNPDA 28 K 28 K 999999 K 999999 K
0.13% DSILOG 26 K 73 K 999999 K 999999 K
0.13% DSIKREM 25 K 73 K 999999 K 999999 K
0.12% DSIROVS 24 K 77 K 999999 K 999999 K
0.11% DSIELTSK 22 K 73 K 999999 K 999999 K
0.11% OPER1 22 K 73 K 999999 K 999999 K
0.10% OPER2 20 K 20 K 999999 K 999999 K
0.09% DSITRACE 18 K 73 K 999999 K 999999 K
0.04% DUIFSSCO 8 K 8 K 999999 K 999999 K
0.04% NTV98BRW 8 K 8 K 999999 K 999999 K
0.03% NTV98 6 K 6 K 999999 K 999999 K

Figure 24. Sample TASKMON Output (Part 1 of 2)

108 IBM Z NetView: Tuning Guide

 0.02% DSIMONIT 4 K 4 K 999999 K 999999 K
0.02% DSIDCBMT 3 K 4 K 999999 K 999999 K
0.02% DSILOGMT 4 K 4 K 999999 K 999999 K
0.02% DSIHLLMT 3 K 4 K 999999 K 999999 K
0.02% CNM01QSD 4 K 59 K 999999 K 999999 K
0.02% DSIQSD4A 4 K 59 K 999999 K 999999 K
0.02% DSIQSD4B 4 K 59 K 999999 K 999999 K
0.02% DSIQSD4C 4 K 59 K 999999 K 999999 K
0.02% DSISTMMT 4 K 4 K 999999 K 999999 K
0.02% SYSOP 3 K 4 K 999999 K 999999 K
0.02% CNMCALRT 4 K 4 K 999999 K 999999 K
0.02% GOPHER 3 K 58 K 999999 K 999999 K
0.02% CAPTAIN 3 K 58 K 999999 K 999999 K
0.02% DSIRQJOB 4 K 4 K 999999 K 999999 K
0.02% DSITIMMT 3 K 4 K 999999 K 999999 K
Severity Index OPID Current Session Cur 24bit Sess 24bit
----DSIGET----- -------- --------- --------- --------- ---------
100.00% AUTO2 20120 K/m 521 K/m 8971 K/m 99 K/m
0.02% NTV98VMT 4 K/m 8 K/m 0 K/m 4 K/m
Severity Index OPID Current Session Cur 24bit Sess 24bit
----DSIFRE----- -------- --------- --------- --------- ---------
100.00% AUTO2 17779 K/m 479 K/m 8977 K/m 99 K/m
0.04% NTV98PPT 7 K/m 27 K/m 0 K/m 6 K/m
Severity Index OPID Current Session Maximum Limit
----MQS In----- -------- --------- --------- --------- ---------
0.91% AUTO2 1 K/m 1 K/m 153 K/m 110 K/m
0.00% NTV98PPT 7 K/m 4 K/m 41 K/m 999999 K/m
0.00% DSILOG 1 K/m 1 K/m 71 K/m 999999 K/m
Severity Index OPID Current Session Maximum Limit
----MQS Out---- -------- --------- --------- --------- ---------
4.35% AUTO2 5 K/m 0 K/m 69 K/m 115 K/m
0.00% NTV98PPT 1 K/m 1 K/m 28 K/m 999999 K/m
0.00% NTV98VMT 1 K/m 2 K/m 54 K/m 999999 K/m
Severity Index OPID Current Session Maximum Limit
------I/O------ -------- --------- --------- --------- ---------
32.45% AUTO2 6490 /m 144 /m 26036 /m 20000 /m
0.00% DSILOG 11 /m 10 /m 395 /m 999999 /m
TASKMON ----- END OF REPORT -----

Figure 25. Sample TASKMON Output (Part 2 of 2)

Note: The output under each group is sorted by the severity index. The Severity Index represents
percentage of the maximum value allowed, or if no limit, the maximum value measured for any task.
The command WINDOW TASKMON * * (TAKE 4 produces a panel that displays the highest four tasks in
each of the measured categories. The WINDOW refresh function key can be used to see updated values.
TASKMON output is color coded based on severity.
For more information on using the TASKMON command and its operands, see the IBM Z NetView
Command Reference Volume 1 (A-N) or the NetView online help.

TASKUTIL Command
The TASKUTIL command displays task performance information, including central processing unit (CPU)
utilization, queue lengths, storage use, and active command lists. This command is for NetView diagnosis
and tuning purposes only.
The TASKUTIL command has three parameters:
TYPE
Specifies the type of NetView task:
ALL
All active NetView tasks. ALL is the default.
AUTO
NetView automation operator station tasks started with the AUTOTASK command. This does not
include operator station tasks (OSTs) or distributed automation tasks (DISTs).
DIST
NetView distributed automation tasks started with the RMTCMD command. This does not include
OSTs or autotasks.

Chapter 10. Additional Tuning Considerations 109

 DST
NetView data services tasks (DSTs). This does not include optional tasks (OPTs).
HCT
NetView hard copy log tasks.
MNT
NetView main task.
NNT
NetView-NetView tasks.
OPT
NetView optional tasks. This does not include DSTs.
OST
NetView operator station tasks. This does not include autotasks or DSTs.
PPT
NetView primary program operator interface task (PPT).
VOST
Virtual Operator Station Tasks (VOSTs). A VOST is created during the ATTACH phase of each full
screen automation request.
TYPE can also specify an individual task name.
DURATION
Specifies the length of the measurement, in seconds, over which utilizations are to be calculated.
Valid values are from 1–60 seconds. The default is 2 seconds.
SORT
Specifies how the output should be sorted. For example, by NAME, TYPE, or CPU percentage (CPUP).
CPUP is the default value.
See the IBM Z NetView Command Reference Volume 1 (A-N) or the NetView online help for a complete
description of the TASKUTIL command and its parameters.

TASKUTIL Command Output

To display CPU utilization and storage used for NetView DSTs, enter the following command:

TASKUTIL TYPE=DST

You receive a response similar to the output in Figure 26 on page 110.

DWO022I
TASKNAME TYPE DPR CPU-TIME N-CPU% S-CPU% MESSAGEQ STORAGE-K CMDLIST
-------- ---- --- ------------ ------ ------ -------- --------- --------
AAUTSKLP DST 249 22019.13 49.02 9.37 0 87521 N/A
BNJDSERV DST 250 4466.25 7.35 1.41 0 357 N/A
DSIELTSK DST 253 4731.99 7.24 1.38 0 31 N/A
DSICRTR DST 251 1362.16 1.97 0.38 0 32 N/A
DSILOG DST 254 624.64 1.40 0.27 0 23 N/A
DSIAMLUT DST 248 1145.74 1.34 0.26 0 26 N/A
AAUTCNMI DST 249 94.44 0.33 0.06 0 463 N/A
CNMTAMEL DST 249 0.36 0.00 0.00 0 49 N/A
CNM01LUC DST 251 306.54 0.00 0.00 0 43 N/A
DSIGDS DST 254 1.89 0.00 0.00 0 46 N/A
DSIHPDST DST 252 2.15 0.00 0.00 0 39 N/A
DSIKREM DST 250 2.15 0.00 0.00 0 549 N/A
DSIROVS DST 251 0.03 0.00 0.00 0 13 N/A
DSISVRT DST 253 0.93 0.00 0.00 0 105 N/A
DSIUDST DST 250 2.59 0.00 0.00 0 14 N/A
DSI6DST DST 251 28.98 0.00 0.00 0 41 N/A
NETVIEW OTHR N/A N/A 0.00 0.00 N/A N/A N/A
NETVIEW SRB N/A 4026.90 5.93 1.13 N/A N/A N/A
NETVIEW TOTL 157 54766.96 100.00 19.11 253 157477 N/A
SYSTEM TOTL N/A N/A N/A 63.70 N/A N/A N/A

Figure 26. TASKUTIL Command Output

110 IBM Z NetView: Tuning Guide

This output was created using the SORT parameter default CPUP and the DURATION default of 2 seconds.
See the IBM Z NetView Command Reference Volume 1 (A-N) for a full explanation of the output fields.
The following events occur when the TASKUTIL command is invoked:
1. CPU time readings are taken for each NetView task. These are the cumulative CPU time values shown
under the heading CPU-TIME.
2. The task processing the TASKUTIL command waits for the amount of time equal to the value of the
DURATION parameter. The task is unable to process commands or messages during this time.
3. When the wait is over, a second set of CPU time readings is taken for each NetView task.
The two fields most important to tuning and diagnosis are N-CPU% (NetView program CPU utilization) and
S-CPU% (system CPU utilization).
The N-CPU% and S-CPU% results reflect utilization over the measurement interval specified by the
DURATION parameter of the TASKUTIL command. For short intervals, such as the default of 2 seconds,
these utilizations are only snapshots, and can be subject to wide variation as the workload of the NetView
program fluctuates. If you use a longer measurement duration, you will get more meaningful utilization
results. The DURATION parameter has a limit of 60 seconds, because suspending a task for a longer
period could cause problems if the task is receiving messages.
It is helpful to understand the N-CPU% and S-CPU% fields and how their values are calculated. The CPU
utilization percentage values are calculated from the two sets of CPU time readings.

Task NetView CPU Utilization

N-CPU% is the task's relative contribution to the CPU utilization of the NetView program, based on a
maximum of 100%. The formula is:

(Task 2nd reading - Task 1st reading)

N-CPU%=------------------------------------------------------- × 100%
(NetView total 2nd reading - NetView total 1st reading)

The numerator represents the task control block (TCB) time used by the task during the measurement.
The denominator represents the total CPU time (TCB time + system request block (SRB) time) used by the
NetView program during the measurement. Multiplying the result by 100% expresses the utilization as a
percentage.

Task System CPU Utilization

S-CPU% is the task's contribution to the total system CPU utilization, based on a maximum of 100%. The
formula is:

(Task 2nd reading - Task 1st reading)

S-CPU%=---------------------------------------------------------x100%
(Measurement duration × Number of online host processors)

The numerator represents the TCB time used by the task during the measurement. The denominator
represents the total CPU time that was available during the measurement (the total capacity of the host
processors). The measurement duration must be expressed in seconds. Multiplying the result by 100%
expresses the utilization as a percentage.

Other CPU Utilization

The NETVIEW OTHR category represents TCB utilization that cannot be attributed to active tasks. The
following formula is used to calculate the value in the NETVIEW OTHR CPU-TIME field is:

OTHR CPU TIME=TOTAL CPU TIME - SRB CPU TIME - (Sum of CPU TIME for each active task)

When the NETVIEW OTHR CPU-TIME value is calculated, it is used in the numerators of the N-CPU% and
S-CPU% formulas to calculate the NETVIEW OTHR N-CPU% and S-CPU% utilization values.

Chapter 10. Additional Tuning Considerations 111

 SRB Utilization
To calculate the NetView SRB N-CPU% and S-CPU% utilization values, two readings of the NetView
address space cumulative SRB time are used in the numerators of the N-CPU% and S-CPU% formulas.

NetView Program Total CPU Utilization

For the NETVIEW TOTL utilization results, two readings of the NetView address space cumulative CPU
time (TCB + SRB) are used in the numerator of the N-CPU% and S-CPU% formulas. For the N-CPU% field,
the result is always 100%. The S-CPU% field for NETVIEW TOTL is the same result as the NetView CPU
% field reported by the RESOURCE command, except that the RESOURCE command uses a 1 second
measurement duration. See “RESOURCE Command” on page 104 for more information.

System Total CPU Utilization

The SYSTEM TOTL utilization is the average processor utilization percentage for all processors currently
online. This result is calculated using the wait time for each of the host processors as follows:

Sum of wait time for all processors during measurement

(1 - ---------------------------------------------------------) × 100%
(Measurement duration × Number of online host processors)

The SYSTEM TOTL CPU utilization is the same result as the TOTAL CPU % field reported by the
RESOURCE command, except that the RESOURCE command uses a 1 second measurement duration.
See “RESOURCE Command” on page 104 for more information.

Calculating Task Utilizations with Two Observations of TASKUTIL

You can calculate task utilization values over longer measurement intervals using two observations of the
TASKUTIL command. You can use the same N-CPU% and S-CPU% formulas to calculate task utilization
over an hour, an 8 hour shift, a day, or a week.
The output in Figure 27 on page 112 is from a TASKUTIL command invocation taken from the same
system as in Figure 26 on page 110, two hours later.

DWO022I
TASKNAME TYPE DPR CPU-TIME N-CPU% S-CPU% MESSAGEQ STORAGE-K CMDLIST
-------- ---- --- ------------ ------ ------ -------- --------- --------
AAUTSKLP DST 249 23907.17 64.62 9.51 0 88844 N/A
DSIELTSK DST 253 5053.53 8.97 1.32 0 31 N/A
DSIAMLUT DST 248 1210.42 2.32 0.34 0 26 N/A
BNJDSERV DST 250 4747.71 1.90 0.28 0 364 N/A
DSICRTR DST 251 1456.56 1.83 0.27 0 32 N/A
DSILOG DST 254 668.85 0.43 0.06 0 23 N/A
AAUTCNMI DST 249 100.61 0.00 0.00 0 463 N/A
CNMTAMEL DST 249 0.36 0.00 0.00 0 49 N/A
CNM01LUC DST 251 329.23 0.00 0.00 0 43 N/A
DSIGDS DST 254 2.06 0.00 0.00 0 46 N/A
DSIHPDST DST 252 2.23 0.00 0.00 0 39 N/A
DSIKREM DST 250 2.15 0.00 0.00 0 549 N/A
DSIROVS DST 251 0.03 0.00 0.00 0 13 N/A
DSISVRT DST 253 0.93 0.00 0.00 0 105 N/A
DSIUDST DST 250 2.59 0.00 0.00 0 14 N/A
DSI6DST DST 251 31.10 0.00 0.00 0 41 N/A
NETVIEW OTHR N/A N/A 0.00 0.00 N/A N/A N/A
NETVIEW SRB N/A 4279.93 6.26 0.92 N/A N/A N/A
NETVIEW TOTL 157 59017.88 100.00 14.72 275 151527 N/A
SYSTEM TOTL N/A N/A N/A 60.51 N/A N/A N/A

Figure 27. Output from TASKUTIL Command, 2 Hours Later

It is evident that AAUTSKLP, the session monitor data services task, is the biggest contributor to NetView
CPU use. Over the 2 hour period, AAUTSKLP used 1888.04 CPU seconds (23907.17 - 22019.13), while
the NetView program as a whole used 4250.92 CPU seconds (59017.88 - 54766.96). For the 2 hour
period, AAUTSKLP's relative contribution to the total CPU utilization according to the N-CPU% formula is:

112 IBM Z NetView: Tuning Guide

 (23907.17 - 22019.13)
--------------------- × 100% = 44.41%
(59017.88 - 54766.96)

Assume that there are six online host processors. AAUTSKLP's contribution to the total system CPU
utilization according to the S-CPU% formula is:

(23907.17 - 22019.13)
--------------------- × 100% = 4.37%
(2 × 60 × 60) × 6

Remember that you need to convert the measurement duration to seconds for these utilization formulas
(2 hours × 60 minutes per hour × 60 seconds per minute = 7200 seconds). This technique makes the
assumption that the task did not stop and restart during the measurement interval.
According to the S-CPU% formula, the contribution to the total system CPU utilization for the 2 hour
period is:

(59017.88 - 54766.96)
--------------------- × 100% = 9.84%
(2 × 60 × 60) × 6

Suggestions for Using TASKUTIL

Suggestions for using TASKUTIL are:
• Although an individual invocation of TASKUTIL provides a valuable picture of the level of activity in the
NetView program, remember that it is only a snapshot. Use multiple invocations of TASKUTIL to spot
trends in task CPU usage and storage growth.
• Consider setting an EVERY timer under an autotask to invoke TASKUTIL every 15 minutes or every hour.
By default, output from the TASKUTIL command (DWO022I) goes to the network log. The TASKUTIL
output can be used as important historical data in diagnosing performance or storage problems. You can
use the BLOG command to browse only messages in the network log generated by the autotask invoking
TASKUTIL.
• You can write applications to invoke TASKUTIL and trap the output for analysis:
– A command list can warn of threshold boundaries for task CPU or storage usage being reached.
– A command list/View panel pair can invoke TASKUTIL and display the data dynamically.
– A REXX command list can reformat the CPU and storage data into a history file using EXECIO.
– A command processor can build SMF records from the TASKUTIL output and write the records to the
external log task (DSIELTSK).
• TASKUTIL output can aid in tuning the NetView program, for example:
– If a command or component is showing constantly high CPU usage, the command or component
should receive focus for tuning or performance improvements.
– If a subset of the autotasks show heavy CPU usage while others have light CPU usage, you could
probably improve performance by redistributing work among the autotasks.
– If the NetView program contributes more to the total system CPU utilization than you would like,
TASKUTIL output could be used to justify component tuning activity or possibly hardware upgrades.
• TASKUTIL output can aid in diagnosis of problems. For example:
– If the task's storage allocation of a task continues to rise, it could indicate that the task is getting
storage, but not freeing it properly.
– If an operator task, autotask, distributed autotask, or NNT shows continually high CPU usage, this
could indicate an endless loop condition in a command list or command. TASKUTIL displays the
active command list for these task types.

Chapter 10. Additional Tuning Considerations 113

 – If an operator task, autotask, distributed autotask, or NNT shows the same command list active, a
message queue build up, and low CPU usage, this could indicate that the command list is stuck in a
WAIT.
– If the message queues for tasks continue to grow during a steady state period when the workload
activity of the NetView program should be fairly uniform, and if the total system CPU utilization is
near 100%, then this could indicate that the NetView program is not getting dispatched frequently
enough to do its work. Continued message queue growth results in continued NetView storage
growth, which can lead to storage abends. If you detect such a condition, consider ending low-
priority CPU-intensive applications to relieve the system CPU constraint. If the NetView program
regularly experiences message queue growth, consider making the MVS dispatching priority for the
NetView address space more favorable.

IBM Z NetView Enterprise Management Agent

Use the IBM Z NetView Enterprise Management Agent to manage your network from the Tivoli Enterprise
Portal. Both sampled and real-time NetView data is available in the Tivoli Enterprise Portal with this
agent. With the IBM Z NetView Enterprise Management Agent and the OMEGAMON® XE performance
agents, you can manage and view availability and performance data for your network from a single
interface.
Data collection for the IBM Z NetView Enterprise Management Agent is done in the NetView address
space. When reviewing the IBM Z NetView Enterprise Management Agent statements in the CNMSTYLE
member, choose the sampling interval values carefully. Here are things to consider:
• If the volume of data that you are collecting for a particular subtower is high, you might want a higher
interval value than for a subtower for which you have less data.
• If you are using OMEGAMON XE for Mainframe Networks and using the NetView cross-product links, you
might want to set your CONNACT and CONINACT subtower interval values to be the same value as the
OMEGAMON XE for Mainframe Networks TCPC collector value.
• DVIPA data collection can involve walking MIBs. This is usually slower than other methods of data
collection.
• Each time a data collection occurs for each subtower, you might see NetView CPU and storage
utilization increase. Use the TASKMON and TASKUTIL commands to monitor CPU and storage for the
autotasks (AUTODCx and AUTOCTx by default) assigned to each subtower.
• If data collection for a subtower does not seem to occur on the designated interval, use the NACTL
LISTINFO or the COLLCTL LISTINFO command (depending on the subtower) to determine how long the
data collection is taking.
See IBM Z NetView Installation: Configuring the NetView Enterprise Management Agent and the IBM
Z NetView Administration Reference for more information about CNMSTYLE statements for the IBM Z
NetView Enterprise Management Agent.
You can also make tuning changes when you configure the IBM Z NetView Enterprise Management
Agent. See the section about tuning considerations in IBM Z NetView Installation: Configuring the NetView
Enterprise Management Agent.

114 IBM Z NetView: Tuning Guide

Chapter 11. Storage Considerations

This section provides techniques for accomplishing the following tasks:

• Estimating storage that is required for the NetView program, RODM, and GMFHS at the host
• Minimizing storage requirements
• Setting REGION sizes

Estimating Storage Usage

A spreadsheet is provided and contains formulas for estimating minimum virtual storage requirements
for the NetView, NetView subsystem, GMFHS, RODM, and Event/Automation Service address spaces, the
RODM data space, and minimum DASD requirements for important NetView, GMFHS, and RODM data
sets.
The spreadsheet, along with a readme file, is contained in the nvstorage_estimator_v6r21.ods file in
the storage_estimator directory on the NetView DVD. The spreadsheet can be used with spreadsheet
programs that support the Lotus® Symphony® ODS format, for example, Lotus Symphony.
If you do not have access to a spreadsheet program, use the formulas in the following tables to estimate
the storage requirements for the NetView program.
If you have access to a spreadsheet program, the following tables can be useful for reference because
they contain additional information not contained in the spreadsheets on how to pick appropriate values
for the input parameters.

Table 3. Base NetView Formulas. All numbers in the results column are in kilobytes (KB).
Default
Name Value Formula Result Description
INITSTART 3 19124 20456 Initial Storage use of the base NetView
program is 19124 KB.
Autotasks 25 _____ * 268 _____ The number of autotasks you expect to have
active on your system. Use the TASKUTIL
command to display the active NetView
tasks on your system. Add the number
of tasks with TYPE=AUTO and TYPE=DIST
(distributed autotasks used by the RMTCMD
command) shown in the TASKUTIL display.
NNTs 5 _____ * 150 _____ The number of NetView-NetView tasks
(NNTs) you expect to have active on your
system. NetView-NetView tasks are started
when an operator from another NetView
domain starts a session to this NetView
domain. For the number of NNTs, use the
TASKUTIL command to display the active
NNTs on your system. Add the number
of tasks with TYPE=NNT shown in the
TASKUTIL display.
BrowseLines 10000 _____ / 50 * 4 _____ Number of lines (or records) in the largest
data set you will browse. Member browse
reads the entire member into storage.

© Copyright IBM Corp. 1997, 2022 115

Table 3. Base NetView Formulas. All numbers in the results column are in kilobytes (KB). (continued)
Default
Name Value Formula Result Description
ExtTrace 0 _____ * 44 _____ Will you be using the external trace function?
Enter 1 if yes or 0 if no. If yes, this adds
in the extra storage required. To determine
if the external trace function is used, use
either the LIST STATUS=TASKS or TASKUTIL
command and see whether task DSITRACE is
active. The default is 1 for yes.
IntTracePgs 0 _____ * 4 _____ How many pages will you be using for
NetView internal trace? The default is 0. If
you start NetView internal trace and do not
specify the SIZE parameter, a default of 250
pages is used. The LIST TRACE command
can be used to determine this value.
SaveRestore 1 _____ * 232 _____ Will you be using the save/restore function?
Enter 1 for yes or 0 for no. If yes, this adds
in the extra storage required. To determine
if the save/restore function is used, use
either the LIST STATUS=TASKS or TASKUTIL
command and see if task DSISVRT is active.
The default is 1 for yes.
StatMon 1 _____ * 145 _____ Will you be using the status monitor? Enter
1 for yes or 0 for no. If yes, this adds in the
extra storage required. To determine if the
status monitor is used, use either the LIST
STATUS=TASKS or TASKUTIL command and
see if a task is active whose name consists
of the five letter NetView domain identifier
followed by the letters ‘VMT’ (for example,
CNM01VMT). The default is 1 for yes.
VTAMnodes 4200 (_____ * 130) / 1024 _____ How many VTAM nodes (major and minor)
will you be monitoring with the status
monitor? A minor node is a uniquely-defined
resource within a major node. A major node
is a set of resources that can be activated
and deactivated as a group. Include all
PUs and LUs defined in the VTAMLST. The
status monitor main panel (domain status
summary) displays this total.
Alias 0 _____ * 92 _____ Will you be using the alias name translation
function? Enter 1 for yes or 0 for no. If
yes, this adds in the extra storage required.
To determine if this function is used, use
either the LIST STATUS=TASKS or TASKUTIL
command and see if task ALIASAPL is active.
The default is 0 for no.

116 IBM Z NetView: Tuning Guide

Table 3. Base NetView Formulas. All numbers in the results column are in kilobytes (KB). (continued)
Default
Name Value Formula Result Description
NtwkProduct 1 _____ * 63 _____ Will you be using the network product
support facility? Enter 1 for yes or 0 for no. If
yes, this adds in the extra storage required.
To determine if this function is used, use
either the LIST STATUS=TASKS or TASKUTIL
command and see if task DSIGDS is active.
The default is 1 for yes.
GdsDSRBs 6 _____ * 1 _____ Specify the number of DSRBs used by task
DSIGDS. This number is the sum of the
DSRBO and DSRBU parameters in DSIPARM
member DSICPINT. The DSRBS DSIGDS
command also displays this number. The
default is 6. This is applicable only if the
network product support facility is used.
CSCF 0 _____ * 38 _____ Will you be using the central site control
facility (CSCF)? CSCF enables you to run
remote online diagnostic tests on 3172 and
3174 devices that support this function.
Enter 1 for yes or 0 for no. If yes, this adds
in the extra storage required. To determine
if this function is used, use either the LIST
STATUS=TASKS or TASKUTIL command and
see if task DSIKREM is active. The default is
0 for no.
ExternalLog 0 _____ * 56 _____ Will you be using the external logging
support facility? Enter 1 for yes or 0 for no. If
yes, this adds in the extra storage required.
To determine if this function is used, use
either the LIST STATUS=TASKS or TASKUTIL
command and see if task DSIELTSK is active.
The default is 0 for no.
InfoMgmt 0 _____ * 1960 _____ Will you be using the Information/
Management Link? Problems can be logged
in the information/management database
from the NetView program. Enter 1 for yes
or 0 for no. If yes, the required extra storage
is added. The default is 0 for no.
MSTransport 1 _____ * 93 _____ Will you be using the management services
(MS) transport? The MS transport is
a programming interface that enables
applications in network nodes to
communicate using conversations over LU
6.2 sessions. MS transport is used for focal
point support, remote bridge support, and
user-written applications. Enter 1 for yes or 0
for no. If yes, this adds in the extra storage
required. To determine if this function is
used, use either the LIST STATUS=TASKS or
TASKUTIL command and see whether task
DSI6DST is active. The default is 1 for yes.

Chapter 11. Storage Considerations 117

Table 3. Base NetView Formulas. All numbers in the results column are in kilobytes (KB). (continued)
Default
Name Value Formula Result Description
MSPartners 20 (_____ * 124) / 1024 _____ Using MS transport, how many partner
LUs will communicate with the NetView
program? Specify the total number of LUs
communicating with applications using MS
transport services. This is applicable only if
you are using the MS transport.
HPTransport 1 _____ * 53 _____ Will you be using the NetView
high performance transport? The high
performance transport offers higher
performance over the MS transport for
applications that require a high transfer rate,
or do not require send confirmations. Enter
1 for yes or 0 for no. If yes, this adds in
the extra storage required. To determine if
this function is used, use either the LIST
STATUS=TASKS or TASKUTIL command and
see if task DSIHPDST is active. The default is
1 for yes.
HPPartners 20 (_____ * 124) / 1024 _____ Using HP transport, how many partner
LUs will communicate with the NetView
program? Specify the total number of LUs
communicating with applications using high
performance transport services. This is
applicable only if you are using the HP
transport.
RMTCMD 1 _____ * 80 _____ Will you be using the RMTCMD command?
RMTCMD allows you to send system,
subsystem, and network commands for
execution under other NetView domains. You
can send only single or multiline messages,
or commands that do not produce full-screen
output. RMTCMD uses high performance
transport services (DSIHPDST). Enter 1 for
yes or 0 for no. If yes, this adds in the
extra storage required. To determine if
this function is used, use either the LIST
STATUS=TASKS or TASKUTIL command and
see if task DSIUDST is active. The default is 1
for yes.
AutoTblSize 1000 (_____ * 200) / 1024 _____ How many IF-THEN and ALWAYS statements
will be in the automation table? The
AUTOCNT TYPE=BOTH STATS=DETAIL
command has the IF-THEN and ALWAYS
statements numbered sequentially for the
MSG and MSU statements. Select the highest
statement number from the MSG and MSU
sections of the detailed report.

118 IBM Z NetView: Tuning Guide

Table 3. Base NetView Formulas. All numbers in the results column are in kilobytes (KB). (continued)
Default
Name Value Formula Result Description
GlobalVars 50 used in next formula What is the total number of global variables
(task and common) you will be using?
Command list global variables are stored in
dictionaries; one task level dictionary per
operator and one common global dictionary
for all operators logged onto the NetView
program. Enter the total number of task and
common global variables to be stored in all
dictionaries. Use the QRYGLOBL command to
determine the number of global variables in
use.
AvgVarSize 40 (GlobalVars_____ * _____ What is the average length (in bytes) of your
(AvgVarSize_____ + 45)) / variables? The length of numeric variables is
1024 4 bytes, while the length of a text variable is
simply the length of the text string in bytes.
Enter the average length for your common
and task global variables. The QRYGLOBL
command can help in determining this value.
UserTasks 0 _____ * 50 _____ How many user-written tasks will you have?
Enter the number of user-written data
services tasks (DSTs) and optional subtasks
(OPTs) for your installation.
PreloadClists 0 used in next formula How many command lists will you preload
with the LOADCL command? Use the MAPCL
command to obtain this information.
AvgClistSize 2000 (PreloadClists_____ * _____ What is the average size (in bytes) of your
AvgClistSize_____) / 1024 preloaded command lists? Use the MAPCL
command to obtain this information.
ResCmdProcs 0 used in next formula How many resident user-written command
procedures will you have? Command
procedures are made resident if the RES=N
parameter on the CMDDEF statement for the
command procedure in member CNMCMD is
omitted.
AvgResSize 2000 (ResCmdProcs_____ * _____ What is the average size (in bytes) of your
AvgResSize_____) / 1024 resident user-written command procedures?
ATFs 0 used in next formula How many user-written automation table
functions (ATFs) will you have? ATFs allow
function calls from the automation table to a
user-written or to a module supplied by IBM,
which can then pass and return information.
When the automation table is loaded, all
ATFs invoked from the table are also loaded.
AvgATFSize 2000 (ATFs_____ * _____ What is the average size (in bytes) of your
AvgATFSize_____) / 1024 user-written ATFs?
Exits 0 used in next formula How many installation exits will you have?
Include both DST (XITxx) and OST (DSIEXxx)
exits.

Chapter 11. Storage Considerations 119

Table 3. Base NetView Formulas. All numbers in the results column are in kilobytes (KB). (continued)
Default
Name Value Formula Result Description
AvgExitSize 2000 (Exits_____ * _____ What is the average size (in bytes) of your
AvgExitSize_____) / 1024 installation exits?

Table 4. Hardware Monitor Formulas. All numbers in the results column are in kilobytes (KB).
Default
Name Value Formula Result Description
HardMon 0 _____ * 433 _____ Will you be using the hardware monitor?
Enter 1 for yes or 0 for no. If yes, this adds
in the extra storage required. To determine
if this function is used, use either the LIST
STATUS=TASKS or TASKUTIL command and
see if task BNJDSERV is active. The default
is 0 for no. The remainder of the parameters
in this table are applicable only if hardware
monitor is used.
ALCACHE 10 (_____ * 500) / 1024 _____ Specify the ALCACHE value from the
CNMSTYLE member. If the ALCACHE value
is set to NONE, use a value of 0. If the
ALCACHE value is set to WRAPCNT, use
a value equal to the alert recording wrap
count, which defaults to 100. The default
value is 10.
TotResources 5000 Used in DASD storage Specify the number of resources in your
formula network that could send alerts to hardware
monitor.
AvgRecords 10 Used in DASD storage Specify the average number of records you
formula expect to have in the hardware monitor
database, per resource. To get a conservative
(high) estimate, you can specify the average
wrap count used for events and statistics
recording. Wrap counts are specified with the
NPDA SWRAP command. The default wrap
count for events and statistics recording is
25.

Table 4. Hardware Monitor Formulas. All numbers in the results column are in kilobytes (KB).
Summary Formula Result
Hardware Monitor Virtual Sum of table values in Results column _____
Storage
Hardware Monitor DASD (500000 + TotResources_____ * (500 + AvgRecords_____ * 700)) / _____
Storage in kilobytes (Kb) 1024

120 IBM Z NetView: Tuning Guide

Table 5. Session Monitor Formulas. All numbers in the results column are in kilobytes (KB).
Default
Name Value Formula Result Description
SessMon 0 _____ * 2327 _____ Will you be using the session monitor? Enter
1 for yes or 0 for no. If yes, this adds in
the extra storage required. To determine if
this function is used, use either the LIST
STATUS=TASKS or TASKUTIL command and
see if task AAUTSKLP is active. The default
is 0 for no. The remainder of the parameters
in this table are applicable only if session
monitor is used.
TsklpDSRBs 10 (6 + roundup(.5 * _____) _____ Specify the number of DSRBOs used by task
+ roundup(.75 * _____)) * AAUTSKLP. This number is specified with
16 the DSRBO parameter in the initialization
member for AAUTSKLP (default AAUPRMLP).
You can also use the DSRBS AAUTSKLP
command to display this number. The default
is 10.
TcnmiDSRBs 11 (_____ + 1) * 16 _____ Specify the total number of DSRBs used by
task AAUTCNMI. This number is the sum of
the DSRBO and DSRBU parameters in the
initialization member for AAUTCNMI (default
AAUCNMTD). You can also use the DSRBS
AAUTCNMI command to display this number.
The default is 11.
RTM 0 Used in session storage _____ Is RTM (response time monitor) active? Enter
formulas 1 for yes or 0 for no. RTM is a NetView
feature available with the 3x74 control unit
to measure response times. To determine if
RTM is used, check the initialization member
for AAUTSKLP (default AAUPRMLP) to see if
RTM=YES is specified. The default is 0 for no.
KEEPRTM 10 Used in session storage Specify the KEEPRTM value in the
formulas initialization member for AAUTSKLP (default
AAUPRMLP), which specifies the number
of collection periods retained for sessions
which have RTM data collected. If you do
not issue the NLDM COLLECT RTM command,
RTM data is only received from the 3x74
when the session ends, and you should
specify a KEEPRTM value of 1. This value is
applicable only if RTM is active. The default is
10.
KEEPSESS 0 Used in DASD storage Specify the KEEPSESS value in the
formulas initialization member for AAUTSKLP (default
AAUPRMLP), which is used to control
DASD session recording and reduce purging
session data from the database. The default
is 0.

Chapter 11. Storage Considerations 121

Table 5. Session Monitor Formulas. All numbers in the results column are in kilobytes (KB). (continued)
Default
Name Value Formula Result Description
SSCP-SSCP 0 (_____ * 370) / 1024 _____ Specify the number of active SSCP-SSCP
sessions. SSCPs (system services control
points) activate, control, and deactivate
network resources. To determine this value,
use the SESSMDIS command.
SSCP-PU 0 (_____ * 370) / 1024 _____ Specify the number of active SSCP-PU
sessions. To determine this value, use the
SESSMDIS command.
SSCP-LU 0 (_____ * 370) / 1024 _____ Specify the number of active SSCP-LU
sessions. To determine this value, use the
SESSMDIS command.
LU-LU 0 (_____ * 260) / 1024 _____ Specify the number of active LU-LU sessions.
To determine this value, use the SESSMDIS
command.
CP-CP 0 (_____ * 370) / 1024 _____ Specify the number of active CP-CP sessions.
To determine this value, use the SESSMDIS
command.
Accounting 0 Accounting_____ * 48 _____ Is Accounting active? Enter 1 for yes or
* (SSCP-SSCP_____ + 0 for no. Session accounting data consists
SSCP-PU_____ + SSCP- of session start and end times, and traffic
LU_____ + LU-LU_____ + counters for the session. If SESSTATS=YES
CP-CP_____) / 1024 is specified in the initialization member
for AAUTSKLP (default AAUPRMLP), session
accounting data is collected for all sessions
processed by the session monitor. The
default is 0 for no.
Availability 0 Availability_____ * 16 _____ Is Availability active? Enter 1 for yes or
* (SSCP-SSCP_____ + 0 for no. Session availability data consists
SSCP-PU_____ + SSCP- of session start and end times for the
LU_____ + LU-LU_____ + session. (Availability data is a subset of the
CP-CP_____) / 1024 data that would be collected if Accounting
was active.) If SESSTATS=AVAIL is specified
in the initialization member for AAUTSKLP
(default AAUPRMLP), session availability
data is collected for all sessions processed
by the session monitor. The default is 0 for
no. This is applicable only if Accounting is not
active.
PctCrossDmn 80 (PctCrossDmn_____ / _____ Specify the percentage (between 0 and 100
100) * LU-LU_____ * 50 / inclusive) of LU-LU sessions that will be
1024 cross-domain. A cross-domain LU-LU session
connects logical units in different domains.
When the two LUs reside in different
domains, each is owned by a different SSCP.
PctCrossNet 10 (PctCrossNet_____ / 100) _____ Specify the percentage (between 0 and
* LU-LU_____ * 150 / 100 inclusive) of LU-LU sessions that will
1024 be cross-network. A cross-network LU-LU
session connects LUs in different networks.

122 IBM Z NetView: Tuning Guide

Table 5. Session Monitor Formulas. All numbers in the results column are in kilobytes (KB). (continued)
Default
Name Value Formula Result Description
PctRTM 50 (PctRTM_____ / 100) * _____ Specify the percentage (between 0 and 100
LU-LU_____ * (80 + inclusive) of LU-LU sessions which have
KEEPRTM_____ * 20) / response time data collected. An LU-LU
1024 session will have response time data if the
terminal is attached to a control unit that
supports RTM and the terminal is owned by
the VTAM host in which the NetView program
is installed. This function is not applicable if
RTM is not active (see the RTM parameter).
PctThruAPPN 50 (PctThruAPPN_____ / _____ Specify the percentage (between 0 and 100
100) * LU-LU_____ * 100 / inclusive) of LU-LU sessions which have
1024 APPN nodes in the session path. Route
Selection Control Vector data is kept for
those LU-LU sessions using a session path
containing APPN nodes.
PctTraced 5 Used in KEEPPIU formula _____ Specify the percentage (between 0 and 100
inclusive) of the total number of sessions
which have PIU trace storage kept.
AvgKEEPPIU 7 (PctTraced_____ / 100) * _____ Specify the average KEEPPIU value for
(96 + KEEPPIU_____ * sessions that have PIU trace data kept.
48) * (SSCP-SSCP_____ A global KEEPPIU value is specified in
+ SSCP-PU_____ + SSCP- AAUPRMLP, and KEEPPIU values can be
LU_____ + LU-LU_____ + specified for individual keep classes on the
CP-CP_____) / 1024 KCLASS statements in the keep member
(default AAUKEEP1).
SessRecorded 25000 Used in DASD storage Specify the total number of sessions that
formulas are recorded to the session monitor VSAM
database during a typical 24-hour day.
To determine this value, you can monitor
the "Sessions Recorded" counter displayed
by the SESSMDIS command over a 24-
hour period. The total number of sessions
recorded will be influenced by the amount of
DASD filtering that you set up.
DaysKept 3 Used in DASD storage Specify the number of days of activity
formulas that are retained in the database when a
purge operation is performed with either the
DBAUTO or NLDM PURGE command. If you
use RESETDB to clear the database, specify a
value of 0.
DaysBetween 7 Used in DASD storage Specify the maximum number of days that
formulas will pass before a database purge operation
is performed or the database is redefined.

Table 5. Session Monitor Formulas. All numbers in the results column are in kilobytes (KB).
Summary Formula Result
Session Monitor Virtual Sum of table values in Results column _____
Storage

Chapter 11. Storage Considerations 123

Table 5. Session Monitor Formulas. All numbers in the results column are in kilobytes (KB). (continued)
Session Monitor DASD maximum of either (1100 * SessRecorded_____ * (DaysKept_____ _____
Storage in kilobytes (Kb) + DaysBetween_____) * 2.1) / 1024 or (1100 * KEEPSESS_____ *
(SSCP-SSCP_____ + SSCP-PU_____ + SSCP-LU_____ + LU-LU_____
+ CP-CP_____)) / 1024

Table 6. TCP Connection Management Monitor Formulas. All numbers in the results column are in kilobytes
(KB).
Default
Name Value Formula Result Description
TCPConn 0 _____ * 83 _____ Will you be using the TCP Connection
Management function? (1=Yes, 0=No)
ActiveConnSess 10000 _____ * 0.0625 _____ How many active TCP connections will be
monitored?

Table 7. VSAM LSR Buffer Storage. Modify the buffer sizes and number of buffers for each size to match the
values you have in CNMSJM01 (DSIZVLSR). All numbers in the results column are in kilobytes (KB).
Buffer Size
Default
and Number Values Formula Result Description
Data Buffers:
CINV, BUFNO 7168, CINV_____ * _____ For each buffer size in the output of the
4 BUFNO_____ VSAMPOOL command, specify the buffer
size (CINV) and number of buffers (BUFNO).
For more information on buffer size and
the number of buffers in specific VSAM
LSR buffer storage configurations, see
“Definitions for LSR and DFR” on page 82 .
CINV, BUFNO 8192, CINV_____ * _____
20 BUFNO_____
CINV, BUFNO 16384, CINV_____ * _____
4 BUFNO_____
CINV, BUFNO 18432, CINV_____ * _____
20 BUFNO_____
CINV, BUFNO 20480, CINV_____ * _____
20 BUFNO_____
CINV, BUFNO 22528, CINV_____ * _____
20 BUFNO_____
CINV, BUFNO 24576, CINV_____ * _____
20 BUFNO_____
Index Buffers:
CINV, BUFNO 512, 3 CINV_____ * _____
BUFNO_____
CINV, BUFNO 1536, CINV_____ * _____
30 BUFNO_____

124 IBM Z NetView: Tuning Guide

Table 7. VSAM LSR Buffer Storage. Modify the buffer sizes and number of buffers for each size to match the
values you have in CNMSJM01 (DSIZVLSR). All numbers in the results column are in kilobytes (KB). (continued)
Buffer Size
Default
and Number Values Formula Result Description
CINV, BUFNO 2048, CINV_____ * _____
30 BUFNO_____
CINV, BUFNO 2560, CINV_____ * _____
30 BUFNO_____
CINV, BUFNO 3072, CINV_____ * _____
10 BUFNO_____
CINV, BUFNO 3584, CINV_____ * _____
30 BUFNO_____
CINV, BUFNO 4096, CINV_____ * _____
30 BUFNO_____
CINV, BUFNO 4608, CINV_____ * _____
30 BUFNO_____

Table 7. VSAM LSR Buffer Storage. Modify the buffer sizes and number of buffers for each size to match the
values you have in CNMSJM01 (DSIZVLSR). All numbers in the results column are in kilobytes (KB).
Summary Formula Result
VSAM LSR Buffer Virtual Sum of table values in Results column/1024 _____
Storage

Table 8. NetView Subsystem Storage. All numbers in the results column are in kilobytes (KB).
Default
Name Value Formula Result Description
buffer z*60 Because of reentrancy requirements, there
must be enough storage for 2 tables to be
where z is the larger of the
loaded at the same time.
two numbers n and m

CRT m*60 Specify the number of lines in the Command

Revision Table.
where m is the number of
lines in the CRT

MRT n*60 Specify the number of lines in the Message

Revision Table.
where n is the number of
lines in the MRT

Chapter 11. Storage Considerations 125

Table 8. NetView Subsystem Storage. All numbers in the results column are in kilobytes (KB). (continued)
Default
Name Value Formula Result Description
PPIreceivers 1 Used in CSA storage Specify the number of program-to-program
formula interface (PPI) receivers that you will have.
PPI receivers are programs that are related
to the NetView program and user-written
application programs to route data such
as generic alerts to the NetView program
for processing by the hardware monitor,
automation facilities, or installation exits.
To determine this value, use the DISPPI
command to display the list of PPI receivers.
The storage for PPI receivers is allocated in
the Common Storage Area (CSA) in 24-bit
storage (below the 16 MB line).
RVAR (256 + 45*n), where n is Specify the number of revision variables
the number of variables. associated with the Message or Command
Revision table.
TotalLimits 1000 (_____ * 880) / 1024 _____ Specify the total of the buffer limits defined
for each PPI receiver. To determine this
value, use the DISPPI command to display
the list of PPI receivers and their buffer
limits.
TracePPIpgs 0 _____ * 4 _____ Specify the number of pages of storage that
you will use for PPI internal trace data. The
storage required for the PPI internal trace
is based on the user-specified size of the
trace table (SIZE parameter on TRACEPPI
command). The default is 10 pages.

Table 8. NetView Subsystem Storage. All numbers in the results column are in kilobytes (KB).
Summary Formula Result
NetView Subsystem Sum of table values in Results column _____
Virtual Storage
NetView Subsystem (3868 + 280 * PPIreceivers_____) / 1024 _____
Storage in CSA below the
line

Table 9. Additional NetView Storage. All numbers in the results column are in kilobytes (KB).
Default
Name Value Formula Result Description
AON 0 _____ * 9692 _____ Will you be using the Automated Operations
Network (AON) feature? Enter 1 for yes or 0
for no. If yes, this adds in the extra storage
required. To determine if this function is
used, use either the LIST STATUS = TASKS
or TASKUTIL command and see if there are
some AUTxxxx or EZLxxxx tasks active. The
default is 0 for no.

126 IBM Z NetView: Tuning Guide

Table 9. Additional NetView Storage. All numbers in the results column are in kilobytes (KB). (continued)
Default
Name Value Formula Result Description
Counters 10 Used in DASD storage Specify the average number of extended
formula statistical counters per loop. The first
counter listed for each date/time is loop
basic counter 2. All subsequent entries
represent the extended statistical counters
you define.
NETVONLY 0 200 or 400® _____ If your command revision table has any
NETVONLY statements, specify the 200
value. If you expect heavy traffic in such
commands, specify the 400 value.
OSTs 5 _____ * (340 + 50 (if _____ How many operators will be concurrently
session monitor active) + logged on to your NetView system? An
58 (if hardware monitor operator station task (OST) is the subtask
active) + 8 (if status that establishes and maintains an online
monitor active) + 14 (if session with the network operator. Use the
TAF used)) TASKUTIL command to display the active
NetView operators on your system. Add the
number of tasks with TYPE=OST shown in
the TASKUTIL display.
Packet Trace 0 3320 + (_______ / _____ Will you be using the packet trace VSAM
1.2048)Minimum value = database to save packet trace data? If yes,
11620 how many packet trace records will be saved
in the VSAM database?
StatFocalPt 0 _____ * 852 _____ Will this host be a status focal point for the
NetView management console? Enter 1 for
yes or 0 for no. If yes, additional storage
is added. Check the initialization member
for CNMTAMEL task specified in member
DSISTASK. If an AMELINIT statement is
enabled, this host is a status focal point. The
default is 0 for no.
TAF 1 Used in OST formula Will your operators be using the terminal
access facility (TAF)? TAF is a facility that
allows a network operator to control several
subsystems from one NetView terminal.
Enter 1 for yes or 0 for no. If yes, additional
storage is added for each OST. The default is
1 for yes.
TEMA 0 _____ * 42440 _____ Will you be using the IBM Z NetView
Enterprise Management Agent? (1=yes,
0=no)

Chapter 11. Storage Considerations 127

Table 9. Additional NetView Storage. All numbers in the results column are in kilobytes (KB). (continued)
Default
Name Value Formula Result Description
VOSTs 0 _____ * 200 _____ How many Virtual OSTs will be concurrently
logged on to your NetView system? A virtual
Also used in NetView
operator station task (VOST) is created as
below the 16MB line
the result of the ATTCH phase during the
private storage calculation
full screen automation process. Use the
TASKUTIL command to display the active
virtual OSTs on your system. Add the number
of tasks with TYPE=VOST shown in the
TASKUTIL display.
Workstat 20 Used in DASD storage Specify the average number of workstations
formula that will be physically attached to each
financial system controller.

Table 9. Additional NetView Storage. All numbers in the results column are in kilobytes (KB).
Summary Formula Result
Additional NetView Virtual Sum of table values in Results column _____
Storage

Table 10. GMFHS Address Space. GMFHS requires that RODM be active. All numbers in the results column are
in kilobytes (KB).
Default
Name Value Formula Result Description
GMFHS 0 _____ * 6780 _____ Will you be using the Graphic Monitor Facility
host subsystem (GMFHS)? Enter 1 for yes
or 0 for no. GMFHS is an advanced network
management system which employs task-
oriented dialogs in an intelligent workstation
providing a graphics base for IBM systems
and network management products. The
default is 0 for no. If you will not be using
GMFHS, the remainder of the parameters in
this table are not applicable.
GmfhsObjects 1000 Used in RODM data space Specify the number of GMFHS-managed
virtual storage formula. objects that you expect to have in RODM.
See Table 14 on page 131.
GmfhsTrcPgs 0 _____ * 4 _____ Specify the number of pages of storage
used for GMFHS internal trace. Trace pages
are used to record GMFHS data for test
and debug purposes. The TRACEPAGES
parameter is specified in the DUIGINIT
member and works in conjunction with
TRACE=ON.

Table 10. GMFHS Address Space. GMFHS requires that RODM be active. All numbers in the results column are
in kilobytes (KB).
Summary Formula Result

128 IBM Z NetView: Tuning Guide

Table 10. GMFHS Address Space. GMFHS requires that RODM be active. All numbers in the results column are
in kilobytes (KB). (continued)
GMFHS Address Space Sum of table values in Results column _____
Virtual Storage above the
16 MB Line

Table 11. Event/Automation Service Address Space. All numbers in the results column are in kilobytes (KB).
Default
Name Value Formula Result Description
Ev/Aut 0 _____ * 5852 _____ Will you be using the Event/Automation
Service (IHSAEVNT)? Enter 1 for yes or 0
for no. The Event/Automation Service serves
as a gateway for event data between the
IBM Z NetView management environment,
the Tivoli management regions managers
and agents that handle Event Integration
Facility (EIF) events, and SNMP managers
and agents. You can use this gateway
function to manage all network events from
the management platform of your choice.
The default is 0 for no.

Table 11. Event/Automation Service Address Space. All numbers in the results column are in kilobytes (KB).
Summary Formula Result
Event/Automation Service Sum of table values in Results column _____
Address Space Virtual
Storage

Table 12. IBM Z NetView Enterprise Management Agent. All numbers in the results column are in kilobytes (KB).
Default
Name Value Formula Result Description
TEMA 0 Used in RTE and Agent _____ Will you be using the IBM Z NetView
formula for Enterprise Management Agent? (1=yes,
0=no)
InstallLibrary 0 Install Library storage _____ Install Library (DASD). See IBM Z NetView
requirements will be Enterprise Management Agent Program
calculated based on Directory for specific Install Library DASD
information in the EMA requirements.
Program Directory
RTE TEMA_____ * 897024 _____ RTE (DASD). Run Time Environment (Full)
- using defaults, with Persistent Data Store
configured.
Agent TEMA_____ * 42800 _____ IBM Z NetView Enterprise Management
Agent address space (Virtual storage).

Chapter 11. Storage Considerations 129

Table 13. RODM Address Space Storage. All numbers in the results column are in kilobytes (KB).
Default
Name Value Formula Result Description
RODM 0 _____ * 49055 _____ Will you be using the Resource Object Data
Manager (RODM)? Enter 1 for yes or 0 for no.
RODM is an object-oriented data store which
supports on-line, real-time collection and
access of data, mainly network and system
management data. RODM is required in order
to use the Graphic Monitor Facility host
subsystem. The default is 0 for no. If you
will not be using RODM, the remainder of the
parameters in this table are not applicable.
ConcurUsers 10 Used in formulas Specify the number of RODM users
connected concurrently. To determine this
value, check member EKGCUST for the
CONCURRENT_USERS value. Allow room
for growth, but do not waste storage
unnecessarily.
AsyncTasks 5 Used in formulas Specify the number of asynchronous RODM
tasks. To determine this value, check
member EKGCUST for the ASYNC_TASKS
value. Allow room for growth but do not
waste storage unnecessarily.
PLI_ISA 40 PLI_ISA_____ * _____ Specify the amount of stack storage in KB
(ConcurUsers_____ + that will be allocated for each RODM user.
AsyncTasks_____) To determine this value, check member
EKGCUST for the PLI_ISA value. This is
storage below the 16 MB line set aside for
PL/I intermodule communication. Use 40 K
for faster run time with minimum wasted
space.
PrimaryHeap 64 PrimaryHeap_____ * _____ Specify the amount of primary PL/I heap
(ConcurUsers_____ + storage in KB that will be allocated
AsyncTasks_____) for each RODM user. To determine
this value, check member EKGCUST for
the PRIMARY_HEAP_SIZE value. This is
temporary storage allocated by RODM
modules for work area. Use 64 K for faster
run time with minimum wasted space.
LogRecords 30000 Used in the following Specify the number of records you want to
DASD storage formula. keep in each log (primary and secondary).

Table 13. RODM Address Space Storage. All numbers in the results column are in kilobytes (KB).
Summary Formula Result
RODM Address Space Sum of table values in Results column _____
Virtual Storage
RODM Storage in CSA 64 * (ConcurUsers_____ + AsyncTasks_____) / 1024 _____
below the line
RODM Log DASD Storage LogRecords_____ / 2 _____
in kilobytes (Kb)

130 IBM Z NetView: Tuning Guide

Table 14. RODM Object Storage. All numbers in the results column are in kilobytes (KB).
Default
Name Value Formula Result Description
UserObjects 0 Used in formula Specify the number of RODM objects created
by user-written RODM applications. To
determine this value, use the RODMUNLD
facility provided with the IBM Z NetView
program with the REPORTONLY=YES option
to get a detailed report of objects residing
in RODM. For prior NetView releases, use
the Create Object API counters in the
STATAPI record. Create a STATAPI record
prior to loading the user objects, and a
second STATAPI record after loading the user
objects. The differences in the Create Object
counters between the two readings can be
used to estimate the total number of user
objects created. See “RODM API Statistics”
on page 74 for more detail on the STATAPI
record.
UserObjSize 3072 (UserObjects_____ * _____ Specify the average size (in bytes) for the
UserObjSize____) / 1024 RODM objects created by user-written RODM
applications. To determine this value, use
cell pool usage information in the STATCELL
record. Create a STATCELL record prior to
loading the user objects, and a second
STATCELL record after loading the user
objects. The differences in cell pool usage
between the two readings can be used to
calculate the total storage used by the user
objects.

Table 14. RODM Object Storage. All numbers in the results column are in kilobytes (KB).
Summary Formula Result
RODM Address Space 13005 _____
Virtual Storage
RODM Data Space Virtual UserObject storage (see previous formula) + 3686 + _____
Storage (GmfhsObjects_____ * 3) + ((Table 9 on page 126 NetView table)
* 3.4)
RODM Checkpoint Data RODM Data Space Virtual Storage for objects + 16384 _____
Set DASD Storage in
kilobytes (KB)
RODM Translation Data RODM Address Space Virtual Storage for objects + 16384 _____
Set DASD Storage in
kilobytes (KB)

Table 15. Minimum Virtual Storage Requirements Summary. All numbers in the Results column are in kilobytes
(KB). To convert values to megabytes (MB), divide by 1024.
Storage Type Table Location/Formula/Amount Result
NetView Address Space See Table 3 on page 115, Table 4 on page 120, Table 5 on page 121, _____
above the 16 MB Line Table 7 on page 124, and Table 9 on page 126.

Chapter 11. Storage Considerations 131

Table 15. Minimum Virtual Storage Requirements Summary. All numbers in the Results column are in kilobytes
(KB). To convert values to megabytes (MB), divide by 1024. (continued)
Storage Type Table Location/Formula/Amount Result
NetView Address Space ((#Autotasks + #NNTs + #UserTasks + #OSTs + #VOSTs) * 2) + 538 _____
below the 16 MB Line
NetView Subsystem See Table 8 on page 125. _____
Address Space above the
16 MB Line
NetView Subsystem 8 _____
Address Space below the
16 MB Line
RODM Address Space Sum of results in Table 13 on page 130 + Address Space object storage _____
above the 16 MB Line (see Table 14 on page 131).
RODM Address Space Environment Dependent _____
below the 16 MB Line
RODM Data Space above Data Space object storage (see Table 14 on page 131). _____
the 16 MB Line
RODM Data Space below N/A _____
the 16 MB Line
GMFHS Address Space 692 plus the value for GmfshTrcPgs that was calculated in Table 10 on _____
above the 16 MB Line page 128 .
GMFHS Address Space 7084 _____
below the 16 MB Line
Event/Automation Service 4800 _____
above the 16 MB Line
Event/Automation Service 2400 _____
below the 16 MB Line
CSA Storage below the 16 See Table 13 on page 130 and Table 8 on page 125. _____
MB Line

Table 16. DASD Storage Conversion Table. Use to convert kilobytes of storage into Cylinders or Blocks.
DASD Type Kilobytes per unit Unit Type
3390 720 Cylinders
3375 384 Cylinders
3330 228 Cylinders
9335 0.5 Blocks
9332 0.5 Blocks
3370 0.5 Blocks
3310 0.5 Blocks

132 IBM Z NetView: Tuning Guide

Table 17. Minimum DASD Storage Requirements Summary. Use the DASD conversion table to convert storage in
KB into blocks or cylinders.
Storage Type Table Location Device Type Formula Result
Session Monitor See Table 5 on _____ Round up (StorageInKb_____ / _____
primary page 121. KbPerUnit_____) and use minimum
of 3
Session Monitor _____ Round up (0.2 * Session Monitor _____
secondary primary result _____)
Hardware See Table 4 on _____ Round up (StorageInKb_____ / _____
Monitor primary page 120. KbPerUnit_____) and use minimum
of 3
Hardware _____ Round up (0.2 * Hardware Monitor _____
Monitor primary result _____)
secondary
TCPCONN See Table 6 on _____ Round up ((StorageInKb_____ * _____
Primary page 124. 92.16) / KbPerUnit_____) and use
minimum of 50 when the TCP
Connection Management function
was selected
TCPCONN _____ Round up (0.125 * TCPCONN _____
Secondary Primary result _____) and use
minimum of 10 when the TCP
Connection Management function
was selected
Save/Restore See Table 3 on _____ Round up (StorageInKb_____ / _____
page 115. KbPerUnit_____) and use minimum
of 2
RODM Log See Table 13 on _____ Round up (StorageInKb_____ / _____
primary page 130. KbPerUnit_____) and use minimum
of 1
RODM Log _____ Same as RODM Log primary _____
secondary
RODM See Table 14 on _____ Round up (Address Space _____
Translation Data page 131. StorageInKb_____ /1024/16) * 16
Set * 1024/KbPerUnit_____
RODM See Table 14 on _____ Round up ( round up (Data Space _____
Checkpoint Data page 131. StorageInKb_____ / 1024 / 16) / 2 )
Set 1 * 16 * 1024 / KbPerUnit_____
RODM _____ Same as RODM Checkpoint Data _____
Checkpoint Data Set 1
Set 2
Canzlog _____ See “Canzlog Archive Storage _____
Archiving Requirements” on page 93.

Region Size
The REGION parameter on the EXEC statement in the startup procedure for a program specifies how
much virtual storage the program is enabled to allocate. Specifying various values for the region size has
the following results (unless your installation changes the default limits in MVS that are supplied by IBM):

Chapter 11. Storage Considerations 133

 • A value equal to 0 MB allows the program to allocate all of the storage it requires below and above the
16 MB line.
• A value greater than 0 MB and less than or equal to 16 MB establishes the size of the private area below
16 MB. The extended region size (above the 16 MB line) is the default value of 32 MB.
• A value greater than 16 MB and less than or equal to 32 MB gives the program all the storage available
below 16 MB. The extended region size is the default value of 32 MB.
• A value greater than 32 MB gives the program all the storage available below 16 MB. The extended
region size is the specified value.
The default REGION size in the NetView startup procedure is 96 MB. Consider changing the REGION size
to 0 MB, which enables the NetView program to allocate the storage it needs and minimizes the risk of
out-of-storage failures. If you do not want to use a value of 0 MB, use no less than 96 MB so that you do
not restrict the use of NetView storage usage below the 16 MB line. Restricting storage below the 16 MB
line does not help other jobs in the system, and could cause out-of-storage failures unnecessarily.

If you want information about... Refer to...

Estimating the amount of virtual storage the “Estimating Storage Usage” on page 115
NetView program requires for your environment
The REGION parameter The appropriate MVS JCL publication
Region size IBM Z NetView Administration Reference

Keeping Track of Virtual Storage and Other System Resource Usage

To collect statistics that help you tune your systems' performance and avoid costly system shutdowns,
periodically check the amount of virtual storage, CPU, and other resources your tasks are using. For
an accurate picture of the resource usage, you probably need to track usage for an entire business
processing cycle. For example, you might want to track usage for a full month to see the usage peaks for
applications and tasks that are not run daily.
To collect statistics on virtual storage, CPU usage, message queuing, and input/output activity for a
system, issue the following timed LOGTSTAT command from a NetView command line:

EVERY 00:30:00,PPT,LOGTSTAT

The example command writes the results to the system monitoring facility (SMF) log. When you have
collected data for the number of days you want, turn off the EVERY command by issuing a PURGE TIMER
command with the timer ID of the original EVERY command. (If you do not know the timer ID, you can find
it by issuing a LIST TIMER command.)
After the statistics are collected in the SMF log, use that data to determine the peak resource usage
for various tasks. Use the NetView function that enables you to issue the DEFAULTS and OVERRIDE
commands to set limits on resource usage for each critical task. When limits are set, the NetView program
sends a warning message when one of the limits is reached or exceeded by a task. You will have time to
react before the task can use enough virtual storage or CPU time to slow down or halt your system. You
can even automate your responses to the warning messages.
To do a quick check of resource usage (for example, for a day or two), use the following command instead
of setting an EVERY timer:

WINDOW TASKMON * *

This command sends the TASKMON output to a window at the operator's console where the command is
issued. It monitors virtual storage, CPU, message queuing, and I/O activity for every task for which you
have set a limit. The results are color-coded, with tasks that are at 90% of their limits shown in red.
To monitor only the virtual storage your tasks are using, issue the following TASKUTIL command each
morning:

134 IBM Z NetView: Tuning Guide

 EVERY 30, TASKUTIL

This command checks the virtual storage usage every 30 minutes for all tasks. You can vary the number of
minutes.
The TASKMON and TASKUTIL outputs provide a baseline from which to compare future data when you
suspect a task might be using an unusual amount of virtual storage. For example, suppose a normal
amount of virtual storage usage for one of your NetView operator tasks is about 2 MB. If the operator
ROLLs from one NetView window to another, the operator task could quickly take up to 20 MB of storage.
Also, the IBM NCP NTuneMon product can use a large amount of virtual storage. If an operator is checking
out two or three NCPs at the same time, it could use in the range of 12–15 MB of virtual storage.
When you have collected data for the specified number of days, turn off the EVERY command by issuing a
PURGE TIMER command with the timer ID of the original EVERY command. (If you do not know the timer
ID, you can find it by issuing a LIST TIMER command.)
Hint: If your TASKUTIL output at 8 a.m. says 70 MB and your REGION size is 75 MB, attempt to determine
immediately what is using the most storage. The amount of virtual storage being used will surely increase
when operators start logging on for the day.
Messages warn you when the NetView region is running out of space. You can automate responses to the
following messages:

BNH162I THE domainid BELOW 16M STORAGE IS nn% USED, mmmK IS LEFT
BNH163I THE domainid ABOVE 16M STORAGE IS nn% USED, mmmK IS LEFT

Minimizing Storage Usage

If you need to save storage, you have several alternatives. When you optimize for storage, you almost
always have a trade-off between storage and host processor utilization.
The constants module (DSICTMOD) enables you to choose whether the first allocation of below-the-line
storage for an individual subpool and size is freed when it is no longer in use. You can use the RESOURCE
command to display the amount of storage in use below-the-line. If the amount of below-the-line storage
used by the NetView program is a problem, set DSICTMOD to free unused storage, although performance
for subsequent uses will be slowed.
If you have fewer than 300 users logged on at any one time, or if you have a large amount of user-written
code that runs in below-the-line storage, choose the option that keeps the first allocation of below-the-
line storage (the default).

Coding RES=N on Command Definition Statements

If you are operating a storage-constrained system, you can save small amounts of storage by coding
RES=N on specific CMDDEF statements in the CNMCMDU initialization member. When a module has
RES=N specified in the CNMCMD initialization member, the module must be dynamically loaded each time
it is used. If you use the module infrequently, the trade-off between saved storage and added processor
time is minimal. For frequently used modules, do not code RES=N because it can result in excessive
system use and I/O activity.
This section lists the command definition (CMDDEF) statements that you can code as RES=N to save
storage.

Alert Network Operations Support

CMDDEF.DSIREGGR.RES=N
CMDDEF.DSILOGGR.RES=N

Coding RES=N on the CMDDEF statements for alert network operations support saves 2 KB.

Chapter 11. Storage Considerations 135

 ALLOCATE, FREE, and LISTA
CMDDEF.ALLOCATER.RES=N
CMDDEF.FREE.RES=N
CMDDEF.LISTA.RES=N

Coding RES=N on the CMDDEF statements for these functions saves 19 KB.

Automated Operations
CMDDEF.AUTOTASK.RES=N
CMDDEF.GENALERT.RES=N
CMDDEF.DEFAULTS.RES=N
CMDDEF.OVERRIDE.RES=N
CMDDEF.GETMSIZE.RES=N,
CMDDEF.GETMTYPE.RES=N
CMDDEF.GETMLINE.RES=N
CMDDEF.PARSEL2R.RES=N
CMDDEF.EXCMD.RES=N

Coding RES=N on the CMDDEF statements for automated operations saves 24 KB.

Automated Operations (MVS)

CMDDEF.MVS.RES=N
CMDDEF.WTO.RES=N
CMDDEF.WTOR.RES=N
CMDDEF.DOM.RES=N
CMDDEF.RELCONID.RES=N
CMDDEF.DISCONID.RES=N

Coding RES=N on the CMDDEF statements for automated operations (MVS) saves 18 KB.

Automation Table Listing

CMDDEF.DSIDTEND.RES=N

Coding RES=N on the CMDDEF statements for this function saves 0.7 KB.

CNM Router

CMDDEF.DSICRUSP.RES=N
CMDDEF.CHNGFP.RES=N
CMDDEF.FPDLCMD.RES=N

Coding RES=N on the CMDDEF statements for the CNM router saves 5 KB.

Commands
CMDDEF.LIST.RES=N
CMDDEF.MSG.RES=N
CMDDEF.HOLD.RES=N
CMDDEF.INPUT.RES=N
CMDDEF.PURGE.RES=N
CMDDEF.ROUTE.RES=N

Coding RES=N on these CMDDEF statements saves 12 KB.

136 IBM Z NetView: Tuning Guide

Cross-Domain CNM Data Transfer

CMDDEF.DSILUITF.RES=N
CMDDEF.DSILCRTR.RES=N
CMDDEF.DSILCSTP.RES=N
CMDDEF.DSILCSF7.RES=N
CMDDEF.DSILCTCU.RES=N
CMDDEF.DSILCNTM.RES=N

Coding RES=N on the CMDDEF statements for data transfer saves 11 KB.

Cross-Domain Logon Facilities

CMDDEF.DSI809A.RES=N

Coding RES=N on the CMDDEF statements for cross-domain logon saves 4 KB.

DCNM Router Function

CMDDEF.DSIFSOLP.RES=N
CMDDEF.DSIFACP.RES=N
CMDDEF.DSIFSCP.RES=N
CMDDEF.DSIFDCP.RES=N
CMDDEF.DSIFRCP.RES=N
CMDDEF.DSIFRMP.RES=N
CMDDEF.DSIFRDP.RES=N
CMDDEF.DSIFRDCP.RES=N
CMDDEF.DSIFDACP.RES=N

Coding RES=N on the CMDDEF statements for the DCNM router saves 13 KB.

External Logging
CMDDEF.DSIELDAT.RES=N

Coding RES=N on the CMDDEF statements for external logging saves 1 KB.

Focal Point Support Over LU 6.2

CMDDEF.DSIFPRCV.RES=N
CMDDEF.DSIFPSND.RES=N

Coding RES=N on the CMDDEF statements for focal point support saves 2 KB.

GLOBALV and Save/Restore Focal Point

CMDDEF.DSIGVRES.RES=N
CMDDEF.AAUDRRES.RES=N
CMDDEF.GLOBALV.RES=N

Coding RES=N on the CMDDEF statements for GLOBALV and save/restore saves 6.5 KB.

HLL and REXX

CMDDEF.WAIT.RES=N
CMDDEF.TRAP.RES=N

Chapter 11. Storage Considerations 137

 Coding RES=N on the CMDDEF statements for HLL and REXX saves 7 KB.

HLL Only
CMDDEF.TIMEP.RES=N
CMDDEF.QUEUE.RES=N
CMDDEF.RID.RES=N

Coding RES=N on the CMDDEF statements for HLL saves 6 KB.

LU 6.2 Transport and Operations Management Support

CMDDEF.DSI6DSCP.RES=N
CMDDEF.DSI6LOGM.RES=N
CMDDEF.DSIOSRCP.RES=N
CMDDEF.DSIOARCP.RES=N
CMDDEF.DSIOURCP.RES=N
CMDDEF.DSIOLGFP.RES=N

Coding RES=N on the CMDDEF statements for LU 6.2 saves 63 KB.

Message Forwarding, Uppercase Translation

CMDDEF.SET.RES=N
CMDDEF.SWITCH.RES=N
CMDDEF.TRACE.RES=N
CMDDEF.UPPER.RES=N

Coding RES=N on the CMDDEF statements for message forwarding saves 10 KB.

NetView Bridge
CMDDEF.RTRQUEUE.RES=N
CMDDEF.TRANRCV.RES=N
CMDDEF.TRANSND.RES=N
CMDDEF.DSINBRSM.RES=N
CMDDEF.DSINBTRM.RES=N

Coding RES=N on the CMDDEF statements for NetView Bridge saves 23 KB.

NetView Bridge Remote Access

CMDDEF.DSINBR62.RES=N
CMDDEF.DSINBRLG.RES=N

Coding RES=N on the CMDDEF statements for NetView Bridge remote access saves 2 KB.

NetView Command List Language and REXX

CMDDEF.DROPCL.RES=N
CMDDEF.LOADCL.RES=N
CMDDEF.MAPCL.RES=N

Coding RES=N on the CMDDEF statements for NetView command list language and REXX saves 9 KB.

NetView management console

CMDDEF.DUIATERM.RES=N

Coding RES=N on the DUIATERM statement for the NetView management console saves 3.5 KB.

138 IBM Z NetView: Tuning Guide

Network Product Support (NPS)
CMDDEF.DISPCMD.RES=N
CMDDEF.CANCMD.RES=N

Service Point

CMDDEF.RUNCMD.RES=N

Other Commands and Command Lists

CMDDEF.AFTER.RES=N
CMDDEF.AT.RES=N
CMDDEF.UNIQUE.RES=N
CMDDEF.EVERY.RES=N

Coding RES=N on the CMDDEF statements for commands and command lists saves 4 KB.

Remote Operations Over the LU 6.2 Transport

CMDDEF.DSIUSNDM.RES=N
CMDDEF.ENDTASK.RES=N

Coding RES=N on the CMDDEF statements for remote operations saves 8 KB.

Save/Restore

CMDDEF.DSITIRTR.RES=N
CMDDEF.RESTORE.RES=N
CMDDEF.AAUDRSRT.RES=N

Coding RES=N on the CMDDEF statements for save/restore saves 15 KB.

Sequential Log

CMDDEF.DSIBSWCP.RES=N
CMDDEF.DSIZBSQW.RES=N

Coding RES=N on the CMDDEF statements for the sequential log saves 5 KB.

Service and Tuning Command Processors

CMDDEF.IDCAMS.RES=N
CMDDEF.SUBMIT.RES=N
CMDDEF.SESSMDIS.RES=N
CMDDEF.DSRBS.RES=N
CMDDEF.LISTCAT.RES=N
CMDDEF.DISPMOD.RES=N
CMDDEF.MSGROUTE.RES=N
CMDDEF.RESOURCE.RES=N
CMDDEF.RESETDB.RES=N
CMDDEF.TASKUTIL.RES=N

Coding RES=N on the CMDDEF statements for service and tuning saves 63 KB.

Chapter 11. Storage Considerations 139

 Session Monitor

CMDDEF.AAUSLGEX.RES=N
CMDDEF.AAUSRTEA.RES=N

Coding RES=N on the CMDDEF statements for the session monitor saves 3 KB.

Status Monitor

CMDDEF.MONIT.RES=N
CMDDEF.STATMON.RES=N
CMDDEF.CLRSTATS.RES=N
CMDDEF.PARSE.RES=N

Coding RES=N on the CMDDEF statements for the status monitor saves 15 KB.

Terminal Access Facility (TAF)

CMDDEF.LISTSESS.RES=N
CMDDEF.ENDSESS.RES=N
CMDDEF.SENDSESS.RES=N
CMDDEF.DSILMEXP.RES=N

Coding RES=N on the CMDDEF statements for TAF saves 8 KB.

140 IBM Z NetView: Tuning Guide

Notices
This information was developed for products and services offered in the U.S.A.
IBM may not offer the products, services, or features discussed in this document in other countries.
Consult your local IBM representative for information on the products and services currently available in
your area. Any reference to an IBM product, program, or service is not intended to state or imply that
only that IBM product, program, or service may be used. Any functionally equivalent product, program, or
service that does not infringe any IBM intellectual property right may be used instead. However, it is the
user's responsibility to evaluate and verify the operation of any non-IBM product, program, or service.
IBM may have patents or pending patent applications covering subject matter described in this
document. The furnishing of this document does not give you any license to these patents. You can
send license inquiries, in writing, to:

IBM Director of Licensing

IBM Corporation
North Castle Drive
Armonk, NY 10504-1785
U.S.A.
For license inquiries regarding double-byte (DBCS) information, contact the IBM Intellectual Property
Department in your country or send inquiries, in writing, to:

Intellectual Property Licensing

Legal and Intellectual Property Law
IBM Japan, Ltd.
19-21, Nihonbashi-Hakozakicho, Chuo-ku
Tokyo 103-8510, Japan

The following paragraph does not apply to the United Kingdom or any other country where such
provisions are inconsistent with local law:
INTERNATIONAL BUSINESS MACHINES CORPORATION PROVIDES THIS PUBLICATION "AS IS"
WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING, BUT NOT LIMITED
TO, THE IMPLIED WARRANTIES OF NON-INFRINGEMENT, MERCHANTABILITY OR FITNESS FOR A
PARTICULAR PURPOSE.
Some states do not allow disclaimer of express or implied warranties in certain transactions, therefore,
this statement might not apply to you.
This information could include technical inaccuracies or typographical errors. Changes are periodically
made to the information herein; these changes will be incorporated in new editions of the publication.
IBM may make improvements and/or changes in the product(s) and/or the program(s) described in this
publication at any time without notice.
Any references in this information to non-IBM Web sites are provided for convenience only and do not in
any manner serve as an endorsement of those Web sites. The materials at those Web sites are not part of
the materials for this IBM product and use of those Web sites is at your own risk.
IBM may use or distribute any of the information you supply in any way it believes appropriate without
incurring any obligation to you.
Licensees of this program who wish to have information about it for the purpose of enabling: (i) the
exchange of information between independently created programs and other programs (including this
one) and (ii) the mutual use of the information which has been exchanged, should contact:

IBM Corporation
2Z4A/101
11400 Burnet Road

© Copyright IBM Corp. 1997, 2022 141

 Austin, TX 78758
U.S.A.

Such information may be available, subject to appropriate terms and conditions, including in some cases
payment of a fee.
The licensed program described in this document and all licensed material available for it are provided by
IBM under terms of the IBM Customer Agreement, IBM International Program License Agreement or any
equivalent agreement between us.
Any performance data contained herein was determined in a controlled environment. Therefore, the
results obtained in other operating environments may vary significantly. Some measurements may have
been made on development-level systems and there is no guarantee that these measurements will be
the same on generally available systems. Furthermore, some measurement may have been estimated
through extrapolation. Actual results may vary. Users of this document should verify the applicable data
for their specific environment.
Information concerning non-IBM products was obtained from the suppliers of those products, their
published announcements or other publicly available sources. IBM has not tested those products and
cannot confirm the accuracy of performance, compatibility or any other claims related to non-IBM
products. Questions on the capabilities of non-IBM products should be addressed to the suppliers of
those products.

Programming Interfaces
This publication documents information that is NOT intended to be used as Programming Interfaces of
IBM Z NetView.

Trademarks
IBM, the IBM logo, and ibm.com® are trademarks or registered trademarks of International Business
Machines Corp., registered in many jurisdictions worldwide. Other product and service names might be
trademarks of IBM or other companies. A current list of IBM trademarks is available on the Web at
"Copyright and trademark information" at http://www.ibm.com/legal/copytrade.shtml .
Adobe is a trademark of Adobe Systems Incorporated in the United States, and/or other countries.
Intel is a trademark or registered trademark of Intel Corporation or its subsidiaries in the United States
and other countries.
Java and all Java-based trademarks and logos are trademarks or registered trademarks of Oracle and/or
its affiliates.
Linux is a registered trademark of Linus Torvalds in the United States, other countries, or both.
Microsoft and Windows are trademarks of Microsoft Corporation in the United States, other countries, or
both.
UNIX is a registered trademark of The Open Group in the United States and other countries.
Other product and service names might be trademarks of IBM or other companies.

Privacy policy considerations

IBM Software products, including software as a service solutions, (“Software Offerings”) may use cookies
or other technologies to collect product usage information, to help improve the end user experience,
to tailor interactions with the end user or for other purposes. In many cases no personally identifiable
information is collected by the Software Offerings. Some of our Software Offerings can help enable you
to collect personally identifiable information. If this Software Offering uses cookies to collect personally
identifiable information, specific information about this offering’s use of cookies is set forth below.
This Software Offering does not use cookies or other technologies to collect personally identifiable
information.

142 IBM Z NetView: Tuning Guide

If the configurations deployed for this Software Offering provide you as customer the ability to collect
personally identifiable information from end users via cookies and other technologies, you should seek
your own legal advice about any laws applicable to such data collection, including any requirements for
notice and consent.
For more information about the use of various technologies, including cookies, for these purposes,
See IBM’s Privacy Policy at http://www.ibm.com/privacy and IBM’s Online Privacy Statement at http://
www.ibm.com/privacy/details the section entitled “Cookies, Web Beacons and Other Technologies”
and the “IBM Software Products and Software-as-a-Service Privacy Statement” at http://www.ibm.com/
software/info/product-privacy.

Notices 143
144 IBM Z NetView: Tuning Guide
Index

Special Characters AUTOCNT command

description 9
? wildcard character 56 detail report 9, 11
.DEFAULT statement 6 overview 2
.NO_ENTRY statement 6 preloaded command lists 24
* wildcard character 56 summary report 11, 12
&CGLOBAL statement 26 AUTODROP command 25
&EXIT statement 25 automating hardware monitor records 13
&PAUSE statement 13 automation
&TGLOBAL statement 26 automation task (AUTOTASK) 13
%INCLUDE statement 7 CMDDEF statement (common) 135
hardware monitor records 13
limiting system messages 5, 7
A MSUs 7, 13
AAU024I message 54, 55 network messages 7
AAUPRMLP member subsystem interface 7
defining SAW buffers 50 system messages 7
defining trace buffer 54 table detail reports 9, 11
keep class member 56 table listings 136
KEEPPIU initialization parameter 58 table summary reports 11, 12
LUCOUNT parameter 66 tuning 5
PURGE parameter 62 automation tasks (autotasks)
selectivity of keep classes 56 &PAUSE statement 13
specifying trace mode 52 GO command 13
tracing gateways 56 multiple 14
tracing LU to LU sessions 52 automation workload
tracing SSCP to SSCP sessions 52 separating from other NetView workloads 15
AAUTSKLP task 50, 54 autotask (automation task)
access method services (AMS) &PAUSE statement 13
LISTCAT 84 GO command 13
REPRO 88 multiple 14
accessibility xii AUTOTBL command 7
accounting information 64 AVAIL parameter 59
ALCACHE statement availability data 59
determining for system 39
using 39 B
alert 37, 46
alert cache 39 background pictures, GMFHS 71
alert network operations support 135 basic tuning information 1
alert recording (AREC) filter 13, 37 BEGIN/END sections, automation table 7
alerts database 13 bisynchronous LU names 56
Alerts-Dynamic panel 39 BLDL search 103
Alerts-Static panel 39 BLDVRP macro 82
alerts, forwarding over LUC session 41 BLOCK parameter 13, 38
ALLOCATE statement 135 BNJ0301 message 13, 37
allocation, data windows 74 BNJ1461 message
ALWAYS statement 7 alert forwarding 41
AMS (access method services) OPER filter 13, 37
LISTCAT 84 BNJDSERV task 39
REPRO 88 BNJMBDST member
APAR OY06920 5 ALCACHE statement 39
API, RODM DSRBO value 44
programming recommendations 79 RATIO (R) statement 46
statistics 74 BNJPNL1 data set 102
AREC filter 13, 37 BNJPNL2 data set 102
assembler language program 23 books
ASSIGN command 9 see publications ix

Index 145
browse, data set 92 CNMSVM10 member 61
buffer cold start 74
allocation 82 COLLECT command 66
control buffer 81 collect parameter 64
data 101 color filter 38
defining 82 command definition (CMDDEF) statement
DFR 81 adding to CNMCMD 99
header 55 list 135, 136
Hiperspace 84 command forwarding 15
I/O 81 command list
lost 54 &PAUSE statement 13
PIU 54 CALL statement 26
pools 82 compiled 26
queue limit 101 converting 26, 31
SAW 50 DROPCL command 24
size 83 executing under an autotask 13
trace 52, 54 extent 103
BUFFERS parameter 83 interpreted 24
BUFNUM parameter 52, 55 library 24
BUFSIZE parameter 52, 55 LOADCL command 24, 103
MAPCL command 24
modifying 24
C name 26
cached DASD device 103 nested 24, 25
calculating CPU utilization 111 preloading frequently used load modules 24
calculating task utilization 112 reducing number and size 24
CALL statement 26 REXX considerations 26
CDINIT 53 subroutine 25
CDRM names 56 system function call 27
cell pool statistics 76 TRACE command 53
CEXEC 26 variable dictionary 35
CGLOBAL 26 command procedure
channel program 81 comment placement 103
checking resource limit 107 common global variables 34
CHKPT command 73, 74 communication between command procedures 26
CI (control interval) compiled 31
size 83 considerations 23
split 88 definition 23
CICS applications 56 HLL command processor 31
CISIZE values, NetView 82 running under different tasks 34
CLEAR parameter 74 running under the same task 34
client programmable workstation 71 subroutine 25
cluster information 85 task global variables 34
CMDDEF statement tuning 23
adding to CNMCMD 99 variables 34
list 135, 136 command processor
CNM router 136 automate messages 7
CNM493l message 9 enhancing performance 35
CNMAUTH statement 100 HLL 31
CNMCMD member 99 initialization 32
CNMCMD residency option, examining 19 PL/I 32
CNMPNL1 data set 102 security 95
CNMS0055 member 61, 101 service and tuning 139
CNMS0080 member 61, 101 commands, performance-related
CNMS7003 (DSIDNMAT) member 46 DSRBS 97
CNMS8003 member 25 LISTCAT 84
CNMSHJ15 member RESOURCE 104
DSICTMOD 101 SESSMDIS 63
sense code filter 61 STATCELL 76
CNMSHM01 member 82 TASKUTIL 109
CNMSHM07 member 61 VSAMPOOL 87
CNMSJM01 member 82 common control blocks 81
CNMSJM10 member 61 common global variables 34
CNMSVM04 member 82 compiled command list 26

146 IBM Z NetView: Tuning Guide

compiled command procedure 31 database (continued)
compiled EXEC (CEXEC) 26 size 62
compiler, REXX 26 VSAM maintenance 88
compiling REXX procedures 26 DBAUTO command
configuration data 50 overview 2
consoles, NetView program 6 VSAM databases 89
constants module, NetView program 101 DCE (data circuit-terminating equipment) 102
CONTINUE 8 DD definition list 103
control area split 88 DEFAULT statement 6
control block DEFAULTS command 30
common 81 deferred write (DFR)
improve access 99 buffer pools 82
search algorithm 67 DSRBO value in BNJMBDST 85
control buffer 81 NetView values 82
control interval (CI) delete operator message (DOM) 7
size 83 dequeue facility 35
split 88 detail reports, automation table 9, 11
conventions DFR
typeface xii buffer pools 82
conversion, REXX command list 26 DSRBO value in BNJMBDST 85
counters 54 NetView values 82
CP-MSU 13, 101 DGROUP option 61
CPU utilization, calculating 111 dialed lines 103
cross-domain directory names, notation xiii
CNM data transfer 137 DISBQL command 101
communications 44 DISKEEP PIU command 57
logon facilities 137 dispatching priority 15
session initiation (CDINIT) 53 DISPPI command
cross-network session initiation 53 overview 2
cross-network session resource allocation 30, 53 usage 101
customization parameters, RODM 79 distributed host 6
DOM (delete operator message) 7
DROPCL command 24
D dropping preloaded command lists 25
D NET command 53 DSIAUTO macro 13
DASD DSICGLOB 8
cached 103 DSICGLOB automation table function 34
DASD parameter 59, 60 DSICLD data set 24, 102
filtering 61 DSICTMOD module
keep-class processing 59 overview 101
session wrap counts 60 sense code filter 61
storage 55, 60 DSIDNMAT (CNMS7003) module 46
DASD I/O 81 DSILIST data set 102
data buffer 101 DSILOG task 82
data circuit-terminating equipment (DCE) 102 DSILUCTD member 100
DATA components 83, 85 DSIMSG 102
data services request block (DSRB) 44, 97 DSINDEF 107
data services task (DST) 44, 97 DSINVGRP 31, 100
data set DSIPARM data set
browse 92 defining SAW buffers 50
closing 81 defining trace buffers 52
log 81 network resource information 107
temporary 103 partitioned data set considerations 102
VSAM 74, 81 selectivity of keep classes 56
data set considerations setting wrap counts 45
NetView control interval size 82 specifying hardware monitor thresholds 46
partitioned 102 DSIPRF data set 103
VSAM LSR and DFR DSIRBS command 97
buffer pool 82 DSIRXFPG module 27
how LSR and DFR work 81 DSIRXUFP module 27
VSAM REPRO command 88 DSIVTAM data set 103
data set member browse 92 DSIZVLSR module 82
data window allocation 74 DSRB (data services request block) 44, 97
database DSRBO parameter 44, 97

Index 147
DSRBS command 97 focal point host
DST (data services task) 44, 97 forwarding alerts 41
DSTINIT statement 97 with distributed hosts 6
FORCE command 60, 81
forwarding alerts 41
E FREE statement 135
E/T (error-to-traffic) threshold 45 full-screen commands 13
EKG1111I error message, RODM 74 function call 27
EKGD001 data set, RODM 74 function packages 27
EKGD002 data set, RODM 74 function routine 7
EKGDWIND member, RODM 74
EKGMAST data set, RODM 74 G
EKGTRAN data set, RODM 74
EKGXRODM JCL, RODM 74 gateway NCP names 56
END statement 7 gateway NCPs 53
end-of-file condition 62 gateway trace 56
enqueue facility 35 generic automation receiver (NVAUTO) 31, 100
environment variables, notation xiii GET request 81
environments, REXX 30 GETC 35
error messages GETMSIZE 26
EKG1111I 74 GETMTYPE 26
error-to-traffic (E/T) ratio threshold GETT 35
RATE statement 46 global trace 8, 52
RATIO statement 46 global variable processing 8, 26
SRATIO command 45 global variables 26, 34
SRFILTER command 45 GLOBALV command 26
ESREC filter 13, 37 GO command 13
ESTAE exit 82 Graphic Monitor Facility, NetView
event counter 54 tuning at host 69
events GROUP statement 103
event record recorded as alert 37
filters for processing 37
logging 46
H
placed in hardware monitor database 81 hardware monitor
processed as subsystem interface function 7 ALCACHE value in BNJMBDST 39
events and statistical recording (ESREC) filter 13, 37 alerts database 13
events and statistics database 13 automating records 13
EXCP (execute channel programs) 86 customizing error-to-traffic ratio 45
execute channel programs (EXCP) 86 database and filters 38
EXIT statement 25 DSRBO value in BNJMBDST 44, 85
exits 82 error-to-traffic threshold 45
extended MCS consoles 6 events and statistics database 13
extent 103 filter 13, 37
external log 54 filter structure 38
external log record 54 HMSTATS 37
issuing SRATIO command 45
F record filtering 13
specifying hardware monitor thresholds 46
filtering tuning 37
DASD 61 HEAP value 32
hardware monitor records 13 helpful hints
HMSTATS command 37 tracking storage use 134
MSUs 13 using TASKMON 134
sense codes 61 using TASKUTIL 134
filters HIER condition 13, 38
AREC 13, 37 high level language (HLL)
COLOR 38 command model statements 137
ESREC 13, 37 command processors 31
hardware monitor 38 preinitialized environments 32
OPER 13, 37 program 23
ROUTE 13, 37 high performance transport 15, 99
SVFILTER command 38 Hiperspace buffers 84
VTAM SAW 50 histogram data 78

148 IBM Z NetView: Tuning Guide

HLLENV command KEEPSESS parameter 62
defining preinitialized environments 32 keyword, MAXSESS 100
overview 2
sample output 33
HMSTATS 37
L
hung terminal 53 language processor environment 30
leased line 15, 103
I limiting system messages 5, 8
linear data sets 73, 74
I/O LINK macro 32
error 103 link speed 71
recording session ends 62 LISTA statement 135
IBM Z NetView Enterprise Management Agent 114 LISTCAT command
IDCAMS EXPORT command 88 access method services (AMS) 84
IDCAMS IMPORT command 88 NetView 84
IDCAMS REPRO command 88 sample output 85
IEBCOPY statement 103 LOAD macro 32
IF-THEN statement 7, 8 LOADCL command
index component 83, 86 automation tasks (autotasks) 14
initialization storage 30 command lists 24, 103
INITMOD statement loading new members 56
defining SAW buffers 50 LOC parameter 82
defining selectivity of keep classes 56 local function package 27
defining trace buffers 54 local shared resource (LSR)
KEEPPIU value 58 buffer pools 82
SESSTATS value 59 definition 81
specifying trace mode 52 DSRBO value in BNJMBDST 85
tracing gateways 56 NetView values 82
tracing LU to LU sessions 52, 53 log data set 81
tracing SSCP to SSCP sessions 53 LOG parameter 59, 82
installation exits, performance 98, 100 log record type 8, RODM 76
interpreted command list 24, 103 logical records 86
interpreted command procedure 31 logical unit (LU)
ISA value 32 number of LUs known by session monitor 66
ISTNMPDS 55 SSCP-LU session 53, 64
ISTNMSDS 51 trace 63
ISTRACON VTAM constants module 55 lookaside hit ratio 87
LSR (local shared resource)
buffer pools 82
J definition 81
job control language (JCL) 74 DSRBO value in BNJMBDST 85
NetView values 82
LU (logical unit)
K number of LUs known by session monitor 66
SSCP-LU session 53, 64
KCLASS AVAIL option 59
trace 63
KCLASS statement 56, 57
LU 6.2 sessions
KEEP classes
command and message forwarding 15
coding 56
communications 72
defining 56
forwarding alerts 41
member 57
NetView constants module 99
naming convention 56
NetView-NetView communication 101
processing for direct access storage device (DASD) 59
overview 99
SAW data 57
LU-LU session
selectivity 56
analyzing network problems 60
specifications 56
configuration data 50
tailoring KEEPPIU values 55
hung terminal 53
use 56
keep class statements 57
keep counts, recommended session monitor PIU 58
protocol problem 53
KEEPMEM member 58, 60
SAW data 50
KEEPPIU option 58
session partners 50
KEEPPIU parameter 53, 55
session status 50, 63
KEEPRTM parameter 64
SESSMDIS command 64
KEEPSESS option 60
TRACELU 52

Index 149
LUC sessions, forwarding alerts over 41 NETCONV statement 71, 72
LUCOUNT parameter 66, 67 NetView
automation 15
consoles used 6
M constants module 101
MACRF values, NetView 82 CPU utilization 111
management services (MS) transport 99 data set member browse 92
managing database size 62 dispatching priority 15
managing preloaded commands 25 function package 27
managing the session monitor database 61 functions for REXX 27
manuals Graphic Monitor Facility 69
see publications ix Graphic Monitor Facility host subsystem (GMFHS) 72
MAPCL command 24 library 24, 26
MAPSESS statement 56, 57 local function 27
MAXSESS keyword 100 primary POI task 35
message problem determination (PD) 15
AAU024I 54, 55 program-to-program interface 101
automated 7 saving storage 135
automation support 5 SESSMDIS command 55, 63
BNJ030I 13, 37 startup JCL 32
BNJ146I startup procedure 103
alert forwarding 41 subsystem job 7
OPER filter 13, 37 system function 27
capture 37 system package 27
CNM493l 9 user function 27
controlling number 5 NetView automation table
detail report 9 %INCLUDE 7
filtering 5 ALWAYS statement 7
forwarding 15 AUTOCNT command 9
summary report 11 automated message 5
suppressed 5 AUTOTBL command 7
traffic 5 BEGIN/END statement 7
message processing facility (MPF) 5, 8 design guidelines 7
messages, limiting system 5, 8 detail report 9
MPF (message processing facility) 5, 8 DSICGLOB 8
MPFLSTxx member 5 efficiency 7
MS transport layer 99 END statement 7
MSUs IF-THEN statement 7, 8
automating 13 message automation 7
detail report 9 message CNM493l 9
filtering 13 MSU automation 7
summary report 11 reducing entries 7
MSUSEG condition 13, 38 related entries 7
multiple autotasks 14 search 5
multiple NetView programs size 7
for separating system and network automation summary report 11
workloads 15 suppressed message 7
multiple processors 14 SYN statement 7
multiple programs 15 NetView programs
multitasking 14 multiple
MVS for system and network automation workloads 15
consoles 6 single
DSRBO values 82 using WLM enclaves 16
NetView workloads
separating automation workload from 15
N NetView-NetView task (NNT)
compared to RMTCMD command 15, 101
naming convention, keep class 56
message forwarding 15, 41
NCCF TRACE options 100
network asset management 102
NCP
network log browse 92
gateway 52, 53
network management vector transport (NMVT) 101
names 56
network message automation 7
trace data 49
NIXL index component 86
nested command lists 24, 25
NLDM RECORD STRGDATA command 54, 63

150 IBM Z NetView: Tuning Guide

NLDM TRACE START command 53 PIU (continued)
NMVT (network management vector transport) 101 keep counts 58
NNT (NetView-NetView task) kept for traced session 55
compared to RMTCMD command 15, 101 trace 64
message forwarding 15, 41 PL/I programs 32
NO_ENTRY statement 6 PLU name 57
nonpersistent session 15, 103 PPT (primary program operator interface task) 35
notation preinitialized environments, PL/I 32
environment variables xiii preloading command lists 24
path names xiii primary program operator interface task (PPT) 35
typeface xiii procedures, REXX 26
number of REXX environments 30 processing, global variables 26
NVAUTO (generic automation receiver) 31, 100 program-to-program interface 101
programmable operator message exchange facility (PMX) 5
programmable workstation
O client 71
OMIT specification 107 connectivity to status focal point 71
online publications storage estimates 70
accessing xi programs, multiple 15
OPER filter 13, 37 protocol problem 53
operator (OPER) filter 13, 37 PU 53, 102
operator request 54 PU names 56
operator station task (OST) 13, 15 publications
OST (operator station task) 13, 15 accessing online xi
OVERRIDE command 30 IBM Z NetView ix
ordering xii
PURGE command 62
P PURGE parameter 62
PURGEDB command 62, 88
panels 103
PARALLEL logmode 99
PARSE EXTERNAL statement 13 Q
parse function 13
PARSE PULL statement 13 QRYGLOBL command
PARSEL2R 26 overview 2
partitioned data set (PDS) 102
partitioned data set directory 103 R
PASS parameter 13
path length, RODM 79 RACMXBUF (SAW buffer limit) 51
path names, notation xiii RACPIULM (PIU data space limit factor) 51, 55
pattern-matching character 56 RACSAWLM (SAW data space limit factor) 55
PAUSE statement 13 RACSAWPK (SAW data space packing factor) 51
PDS (partitioned data set) 102 RECORD command 66
performance considerations record filtering 13
concatenating the DSICLD library 19 region size 133
DDF tree and panel, customizing 20 RELOAD command 56
NetView 19 remote operations over LU 6.2 transport 139
node automation, switching 20 reorganizing database
operations 20 recovering space lost 88
reordering the automation table 19 session monitor 62
performance, enhancing 35, 95 REPRO command 88
persistent session request units (RUs) 97
LUC sessions 103 RESETDB command 62, 88
status focal point 15 RESIDENT entries 26
physical unit (PU) 102 RESOURCE command 104
PIU Resource Object Data Manager (RODM)
buffers API statistics 74
allocation 54 checkpoint guidelines
estimate fullness 54 EKGD001 data set 74
ratio 65 EKGD002 data set 74
size 54, 65 EKGMAST data set 74
storage 54 EKGTRAN data set 74
count 54 customization parameters 79
data space limit factor (RACPIULM) 55 log record type 8 76

Index 151
Resource Object Data Manager (RODM) (continued) SAW (session awareness) data (continued)
path length 79 definition 49
programming recommendations 79 description 50
storage manager 78 discarded 57
storage requirements 130 filtering 50
resource pool 82 filtering by NetView program 50
resource status collector filtering by VTAM 50
resynchronization 71 keep classes 57
response time monitor (RTM) 64 keeping selectively 50
REXX LU-LU session 50, 64
CALL statement 26 non-RTM data 57
command list conversion 26 notifications 50
compiler 26 parameter 63
considerations 26 RTM data 57
environments 30 session partners 50
interpreter 27 session PIU 49
nested command list 26 session status 50, 63
parse function 13 SESSMDIS command 50, 63
storage considerations 30 size of SAW buffers 50
system function call 27 SSCP-LU session 50, 64
RMTCMD command SSCP-PU session 50, 64
command and message forwarding 15, 101 SSCP-SSCP session 50, 64
data buffers 99 storage required per session 50
RNAA 53 VTAM SAW filter 50, 56
RODM (Resource Object Data Manager) SAW parameter 57
API statistics 74 SDLC (synchronous data link control) 72
checkpoint guidelines SDOMAIN command 100
EKGD001 data set 74 search algorithm 67
EKGD002 data set 74 search sequence 27
EKGMAST data set 74 search table 67
EKGTRAN data set 74 security 95
customization parameters 79 security scenarios 79
log record type 8 76 selective tracing 50, 52
path length 79 sense code filtering 61
programming recommendations 79 separating system and network automation workloads
storage manager 78 using multiple NetView programs 15
storage requirements 130 separating the automation workload from other NetView
route (ROUTE) filter 13, 37 workloads 15
ROUTE command 101 sequential log command model statements 139
route selection control vector (RSCV) 65 server-client configurations 72
RSCV 65 service point command model statements 139
RTM (response time monitor) 64 service request block (SRB) 104
run-time library 32 session
RUs (request units) 97 accounting 53, 64
RUSIZES 99 counts 64
cross-domain 50
cross-network 50
S ends 62, 65
samples library 82 initiation 53
Save/Restore NCP trace data 49
command model statements 139 partner 50
processing 35 partner name 56
VSAM data set 35 PIUs 49, 55
SAVEC function 35 recording 60
SAVET function 35 resource allocation 53
SAW (session awareness) data same domain 50
allocating SAW buffers starts 56, 65
MVS/ESA systems 51 status 50, 63
buffer trace data 49
ratio 65 session awareness (SAW) data
size 65 allocating buffers 50
configuration data 50 buffer ratio
data space limit factor (RACSAWLM) 51 ratio 50
data space packing factor (RACSAWPK) 51 configuration data 50

152 IBM Z NetView: Tuning Guide

session awareness (SAW) data (continued) session monitor (continued)
definition 49 warm start 49
description 50 work load traffic counters 65
discarded 57 session status 50
filtering 50 session-end notification 54
filtering by VTAM 50 sessions, persistent and nonpersistent 103
keep classes 56 SESSMDIS
LU-LU session 50, 64 monitoring number of SAW notifications 50
non-RTM data 57 session monitor statistics 63
notifications 50 storage allocated for PIU trace data 55
RTM data 57 tuning PIU buffer allocation 54
session partners 50 SESSTATS parameter
session PIU 55 accounting and availability 53, 59
session status 50 session and storage information panel 63, 64
SESSMDIS command 50, 63 set domain (SDOMAIN) command 100
size of SAW buffers 54 set recording filter (SRF) command 46
SSCP-LU session 50, 64 SETBQL command 101
SSCP-PU session 50, 64 SETCV 53
SSCP-SSCP session 50, 64 SHRPOOL parameter 82
storage required per session 50 single NetView programs
VTAM SAW filter 50, 56 using WLM enclaves 16
session block 67 SLU names 57
session initiation (INIT) 53 SMDR (session monitor data recording) command 62
session monitor SNA LU names 56
accounting 49, 64 SNI session 67
COLLECT and RECORD commands 66 solicited DSRBOs 97
command model statements 140 specific trace 52
data recording (SMDR) command 62 SRATIO command 45
database SRB 104
managing 61 SRFILTER (SRF) command
DISKEEP PIU command 57 disabling network resources 45
failure 52 setting hardware monitor filters 13, 37
global trace mode 52 SSCP session 53
initialization 56 SSCP trace 63
keep class SSCP-LU session
DASD keep-class processing 59 configuration data 50
recommend PIU keep counts 58 keep class statements 57
SAW data 56 SAW data 50
KEEPPIU 58 session initiation (INIT) 53
LUCOUNT parameter 66 session partners 50
NCP trace data 49 session status 50
PIU 49 SESSMDIS command 64
processing 50 TRACESC 52
RELOAD command 56 VTAM application 53
reorganizing database 62 SSCP-PU session
RTM (response time monitor) 49 configuration data 50
SAW (session awareness) data 50 cross-network session resource allocation 53
selective tracing 50, 52 gateway NCPs 53
session and storage information panel 63 SAW data 50
specific trace mode 52 session partners 50
storage usage 64 session status 50
trace data SESSMDIS command 64
gateway 56 TRACESC 52
global 52 SSCP-SSCP session
NCP 49 configuration data 50
PIU 54 cross-domain session initiation 53
recording 52 SAW data 50
selective tracing 52 session failure 53
specific 52 session partners 50
summarized 52 session status 50
viewing 52 SESSMDIS command 64
trace mode 52 TRACESC 52
traffic data 54 startup JCL 32
tuning 49 STATAPI parameter

Index 153
STATAPI parameter (continued) TRACE command specification 52, 53
generating RODM statistics 74 trace options 100
overview 3 TRACE START command 53
STATCELL parameter 76 trace, gateway 56
statistics 37, 76 TRACEGW 56
STATOPT filtering 107 TRACELU 52
status focal point TRACEPPI command 102
workstation connectivity 71 traces, global and specific 52
status monitor TRACESC 52
command model statements 140 TSO applications 57
resource limit 107 TSO/E
STEPLIB DD statement 32, 107 ARXANCHR environment 31
STOP FORCE command 81 IRXANCHR environment 31
storage TSO/E local 27
considerations 30, 135 TSO/E system 27
manager, RODM 78 TSO/E user 27
programmable workstation estimates 70 TSO/E ARXANCHR environment 31
RODM requirements 130 TSO/E IRXANCHR environment 31
storage usage tuning
tracking 134 additional considerations 91
STRNO parameter 83 automated operations 5
subarea dial overhead 15 basic information 1
subpool 54 command procedures 23
subroutines 25 definition 1
subsystem interface 5 hardware monitor 37
subsystem job 7 LU 6.2 transport 99
summary reports, automation table 11, 12 NetView Graphic Monitor Facility
SVFILTER command 38 host 72
SWITCH command 81 workstation 69
switched line 15, 103 RODM 73
SWRAP command 44 session monitor 49
SYN statement 7 status monitor 91
synchronous data link control (SDLC) 72 VSAM 81
system tuning considerations
command 7 concatenating the DSICLD library 19
CPU utilization 111 DDF tree and panel, customizing 20
function call 27 NetView 19
function package 27 node automation, switching 20
library 26 operations 20
message 5, 7 reordering the automation table 19
message traffic 5 typeface conventions xii
system function call 27
U
T
unsolicited DSRBUs 97
task control block (TCB) 104 usage reports, automation table
task global variables 34 detail report 9
task utilization, calculating 112 summary report 11
TASKMON 134 user function package 27
TASKUTIL user group, NetView xii
command output 110 user groups
command processor 109, 139 NetView xii
diagnosis techniques 113
suggestions for using 113
tuning techniques 113
V
TCB 104 variable dictionary 35
temporary VIO data set 103 variables, common global 34
terminal access facility (TAF) 140 variables, notation for xiii
TGLOBAL 26 VIO data set 103
time-out facility 13 virtual I/O (VIO) data set 103
Tivoli Software Information Center xi virtual storage access method (VSAM)
token-ring connections 72 BLDVRP macro 82
TRACE (NCCF) options 100 buffer allocation 82
TRACE command 53

154 IBM Z NetView: Tuning Guide

virtual storage access method (VSAM) (continued) warm start
CI size 81 RODM 73, 74
data set 81 session monitor 49
linear data sets (LDSs) 73, 74 wildcard characters 56
local shared resources (LSR) and deferred write (DFR) window allocation 74
81 WLM (Work Load Manager) 16
record queue 65 Work Load Manager enclaves
resource pool 82 for separating system and network automation
session recording 65 workloads 16
tuning 81 workload, automation
VSAMPOOL command 84 separating from other NetView workloads 15
virtual storage, tracking 134 wrap counts 44, 60
virtual storage, unused 67 wrap counts, initialization 45
virtual telecommunications access method (VTAM) write-to-operator (WTO) command 7
application 53 write-to-operator with reply (WTOR) command 7
constants module (ISTRACON) 55 WRTBFR macro 81
D NET command 53 WTO (write-to-operator) command 7
display command 53 WTOR (write-to-operator with reply) command 7
MVS/ESA considerations 51
naming conventions 56
PIU buffer 54
X
private storage 50 XCTL macro 32, 107
SAW filter 50
sessions 49
vital product data (VPD) 102
VPD (vital product data) 102
VSAM
BLDVRP macro 82
buffer allocation 82
CI size 81
data set 81
database maintenance 88
I/O 81
linear data sets (LDSs) 73, 74
local shared resources (LSR) and deferred write (DFR)
81
record queue 65
REPRO command 88
resource pool 82
session recording 65
tuning 81
VSAMPOOL command 84
VSAM ACB options 85
VSAM REPRO command 88
VSAMPOOL command
buffer pool 84
overview 3
usage 87
VTAM
application 53
constants module (ISTRACON) 55
D NET command 53
display command 53
MVS/ESA considerations 51
naming conventions 56
PIU buffer 54
private storage 50
SAW filter 50, 56
sessions 49
VTAMLST 107

W
WAIT FOR OPINPUT 13

Index 155
156 IBM Z NetView: Tuning Guide
IBM®

SC27-2874-06