Autonomic Computing

Mouzzam Hussain
• BSCS
• Foundation University `07
• [email protected]

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Content
Introduction
History
Vision Of Autonomic Computing
Specialties, Properties, Concept
Autonomic Computing Tree
Dependability
The Control Loop
The PMAC Architecture
The Self Let Approach
Application & Challenges
M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Introduction
Imagine about the Central Nervous System for a moment.
It’s as if the autonomic nervous system says to you, Don ’t
think about it—no need to. I’ve got it all covered.
Autonomic computing attempts to intervene in computing
systems in a similar fashion as its biological counterpart.
Autonomic Computing helps to address complexity by
using technology to manage technology
The term autonomic computing was first used by IBM in
2001 to describe computing systems that are said to be
self-managing [Kephart and Chess].

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 History
The concepts behind self-management* were not entirely new
to IBM’s autonomic computing initiative.
One of the notable early self-managing projects were initiated
by DARPA for a military application in 1997,2001.
Situational Awareness System1 (SAS) SUO program.
DASADA2 (Dynamic Assembly for System Adaptability,
Dependability and Assurance) enabling mission critical
systems to meet high assurance, dependability, and
adaptability requirements.

Essentially, it deals with the complexity of large distributed

software systems, a goal not dissimilar to IBM’s autonomic
computing initiative.
M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 H istory C on td …

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Why Autonomic Computing?
Computing system’s complexity appears to be approaching the
limits of human capability yet the march toward increased
interconnectivity and integration rushes ahead unabated.
This march could turn the dream of pervasive computing—trillions
of computing devices connected to the Internet—into a
nightmare.
Programming language innovations have extended the size and
complexity of systems that architects can design, but relying
solely on further innovations in programming methods will not
get us through the present complexity crisis.
In the keynote address, Paul Horn deliberately chose a
term with a biological connotation.

Paul Horn senior vice president of research IBM @

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Why AC Contd…
The main reason for large blue-chip companies, like
IBM, being interested in autonomic computing is
the need to reduce the cost and complexity of
owning and operating an IT infrastructure .
The aim is to allow administrators to specify high-
level policies that define the goals of the autonomic
system, and let the system manage itself to
accomplish these goals.
At present, system administrators must tweak
hundreds of settings and often spend weeks before
getting a system to run optimally.
Autonomic systems are also faster at adapting to
changes to the environment, e.g. by distributing
2443/FUIEMS/BSCS-07
its
M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 Fe b - 0 4 - 2 0 1 0
 Vision Of AC
Systems manage themselves according to an
administrator’s goals.
New components integrate as effortlessly as a new
cell establishes itself in the human body.
These ideas are not science fiction, but elements of
the grand challenge to create self-managing
computing systems.

Jeffrey O.Kephart, David M.Chess

IBM Thomas J.Watson Research Center

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 what specifically are they?
Such a high-level system could be described as possessing
at least eight key elements or characteristics.

1. “know itself ” and comprise components that also possess a system

identity.
2. configure and reconfigure itself under varying and unpredictable
conditions.
3. never settle for the status quo — it always looks for ways to optimize
its workings.
4. must perform something akin to healing
5. must be an expert in self-protection.
6. knows its environment and the context surrounding its activity, and
acts accordingly.
7. cannot exist in a hermetic environment.
8. anticipate the optimized resources needed while keeping its
complexity hidden.

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Comcepts Of AC

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Concepts of AC Contd

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Autonomic Computing Tree

Fig u re 2 A u to n o m ic C o m p u tin g Tre e

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 AC & Dependability

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 The Control Loop
• A basic concept that shall be applied in
Autonomic Systems are closed control
loops.
•
• A closed control loop in a self-managing
system monitors some resource (software
or hardware component) and
autonomously tries to keep its parameters
within a desired range.

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 THE MAPE-K AUTONOMIC
LOOP
To achieve autonomic computing, IBM has suggested a
reference model for autonomic control loops [IBM 2003],
which is called the MAPE-K (Monitor, Analyse,Plan,
Execute, Knowledge) loop.
This model is being used more and more to communicate
the architectural aspects of autonomic systems.
Likewise it is a clear way to identify and classify much of
the work that is being carried out in the field.
In the MAPE-K autonomic loop, the managed element
represents any software or hardware resource that is
given autonomic behavior by coupling it with an
autonomic manager.
managed element can be a web server or database, a
specific software component in an application

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 THE MAPE-K LOOP Contd..

IB M ’ s M A P E - K ( Monitor , Analyse , Plan , Execute ,

K n o w le d g e ) re fe re n ce m o d e l for a u to n o m ic co n tro l lo o p s.

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Knowledge

§K n o w le d g e Typ e s
•S o lu tio n To p o lo g y K n o w le d g e
•Pro b le m D e te rm in a tio n K n o w le d g e
•Po licy K n o w le d g e
•
§Tw o M e ch a n ism s to o b ta in kn o w le d g e
•E ffe cto r In te rfa ce
•M o n ito r Pa rt

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Policy
§Po licy
•A ctio n p o licy: If ( C o n d itio n ) T H E N
( Action )
•G o a l p o licy: d ire ctio n
•U tility fu n ctio n p o licy: Prio rity

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Autonomic Element

§ AEs are responsible for:

Managing their own behavior
Interacting with other autonomic
elements
§ AEs contain:
one autonomic manager
Zero or more managed element
(s)

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Autonomic Manager

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Managed Element

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Interaction b/w Components Interaction Between
Components

Interfaces
Monitoring and testing
interfaces
Lifecycle interfaces
Policy interfaces
Negotiation interfaces

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Structure of AE’s

Figure3 . Structure of an autonomic element. Elements

interact with other
elements and with human programmers via 2443/FUIEMS/BSCS-07
their autonomic
M Mo ou uzza
managers.
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 Fe b - 0 4 - 2 0 1 0
 Example

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Autonomic System

An autonomic system is a system that operates and

serves its purpose by managing itself without external
intervention

How to build a system with autonomic behavior?

1. A collection of autonomic elements that implement the
desired function.
2. Additional autonomic elements to implement system
functions that enable the needed system level
behaviors.
3. Design patterns for system SELF- CHOP.

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Autonomic System Contd..

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 SELF– CHOP

M o u zza m H u ssa in 2443/FUIEMS/BSCS-07

Fe b - 0 4 - 2 0 1 0
 Design Patterns
Self-Configuring
§ Goal driven self-Assembly
• Initialization
• Registration
Self-Healing
§ Self-regenerating cluster
§ Required interfaces
• Sending state
• Receiving state,
• Querying planned outage (Availability management)

M Mo ou uzza
zzam m HHuussa
ssaiinn
• Scheduling planned outages (Availability
2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Design Patterns Contd..
Self-optimization
§ Market-based control (buyer &seller)
§ Resource arbiter
§ Required interfaces
• Query service
• Query service level bounds
• Requesting a service level
Self-Protecting
• Some aspects are similar to self-healing
• Prevention: Policy-based management
zzam m •
M Mo ou uzza Security
HHuussa
ssa iinn policy 2 4 4 3 /FU IE M S /B S C S -0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 PMAC–Architecture AC
Policy Management for Autonomic Computing (PMAC)*
Purpose: Providing a Policy management infrastructure
Automating what administrators do today
Administrators follow written policies
With autonomic, autonomic managers follow machine-readable
policy
Autonomic Manager – Selects policies, evaluates policies, and
provides decisions to the managed element in order to manage its
behavior
Using Autonomic Computing Policy Language(ACPL) as common
policy language
ACPL contains 4 tuples: Scope, Condition, Business value, Decision
Scope represents managed elements, Business value is the
decision priority
Decision can be Actions, Configuration Profiles and Results
M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 PMAC - Architecture

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 PMAC – Example
Consider the goal policy
– Scope: Company A’s on-line ordering system
– Condition: During business hours
– Business value: 100
– Decision: 2-second average response time
In this case the Managed element is an on-line
ordering system
Autonomic Manager makes the decision by
– Monitoring data coming from the online ordering
system
– Analyzing the gathered data using conditions
(business hours?)
– Planing and executing based on2443/FUIEMS/BSCS-07
the previous
M o u zza mHussain
Mouzzam H u ssa in 2 4 4 3 / FU IE M S / B S C S - 0 7 Feb-04-2010
 SELFLET CONCEPTUAL
MODEL
SelfLet is a self-sufficient piece of software which
is situated in some kind of logical or physical
network, where it can interact and
communicate with other SelfLets.
SelfLets require or offer some kind of services to
other SelfLets in order to comply with their
specified internal behaviour
they can, however, dynamically modify and
adapt this behaviour in reaction to changes in
their internal state or in the environment, in
order to accomplish their high-level goals.

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 An Infrastructure for Autonomic System Development:

the SelfLet Approach

Figure 1: The conceptual model of a SelfLet.

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 SelfLets to SelfLets Interactions

SelfLets interact with their similars through two kinds of

messages, application-level messages and negotiation-
level
Negotiation-level messages are used by SelfLets to
exchange data about Goals
A SelfLet has a Needed Goal (NG) when it cannot perform
on its own some kind of function required by its
Behaviour, and thus it has to resort to the help of other
SelfLets.
vice versa, a SelfLet has an Achievable Goal (AG) when it
can provide some kind of service to “SelfLets in need”.
what is an AG for some SelfLets will be a NG for others:
negotiation-level messages are thus used to “match” NGs
to AGs
M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 SelfLet Approach Contd..

Figure 2: The internal architecture of a

SelfLet
M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
THE SELFLET ARCHITECTURE AND
PROTOTYPE
The Behaviour Manager is responsible for the actual execution
of the SelfLet Behaviours, which typically consist in a
workflow of calls to local Abilities or remote Goals.
The Abilities of a SelfLet are managed by the Ability Execution
Environment, which not only installs, stores and uninstalls
them, but also executes them after receiving either local or
remote requests.
The Negotiation Manager is used to interact with other
SelfLets by publishing the availability of certain Goals or by
requesting one if needed
The communication is performed via the Message Handler
which is the access point for incoming and outgoing
messages.
The Autonomic Manager is perhaps the most complex
component, as it devoted to execute the autonomic
characteristics of the SelfLet. It monitors all the other
components and the messages they exchange, both among
zzamthem
M Mo ou uzza ssaiinnand with other SelfLets.
m HHuussa 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Applications
Autonomic computing promises to simplify the management of
computing systems. But that capability will provide the basis for
much more: from seamless e-sourcing and Grid Computing, to
dynamic e-business and the ability to translate business decisions
that managers make to the IT processes and policies that make
those decisions a reality.
IBM’s Dr Kumar says “One of the first applications of autonomic
computing is e-sourcing”. E-sourcing is the ability to deliver IT as a
utility, when you need it, in the amount you must have to
accomplish the task at hand.
Autonomic computing will create huge opportunities for these kinds
of services, feels Dr Kumar.
Other applications include server load balancing, process allocation,
monitoring power supply, automatic updating of software and
drivers, pre-failure warning, memory error-correction, automated
system backup and recovery, etc.
M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Challenges
Analysts say the days of widespread autonomic computing
usage are still way off. However, we are beginning to
see elements of it in business systems. Meanwhile, the
challenges faced in developing such systems are mainly
those dealing with the management of complex systems
operating in heterogeneous environments.
The other challenge is to convince customers that
autonomic computing actually simplifies systems
management and can cut costs.
IT managers and administrators may be reluctant to give
up control of the systems they manage.

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Availability
Truly autonomic systems are years away, although in the nearer
term, autonomic functionality will appear in servers, storage and
software. For instance, IBM's z900 eServers have a self-managing
operating system known as Intelligent Resource Director (IRD).

IBM z900 e Servers.

M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 Autonomic Computing Today
The ideas behind autonomic computing are not new.
In fact, it is possible to find some aspects of
autonomic computing already in today’s software
products .
Windows XP optimises its user interface (UI) by
creating a list of most often used programs in the
start menu. Thus, it is self-configuring in that it
adapts the UI to the behaviour of the user
It can also download and install new critical updates
without user intervention, sometimes without
restarting the system. Therefore, it also exhibits
basic self-healing properties.
DHCP and DNS services allow devices to self-
M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
Self- organization within  autonomic system

§
§
§

M o u zza m H u ssa in 2443/FUIEMS/BSCS-07

Fe b - 0 4 - 2 0 1 0
 IBM affiliated Research Work*
External Projects:
Berkeley University: OceanStore
Berkeley University: Recovery-Oriented Computing
Carnegie Mellon University: Self-securing Storage
Columbia University: Autonomizing Legacy Systems
Cornell University: Astrolabe
Georgia Institute of Technology: Qfabric

*
IBM Research hprojects:
ttp :// w w w . re se a rch . ib m . co m / a u to n o m ic / re se a rch / p ro je c
M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
 References
http://www.research.ibm.com/autonomic/research
ACM: A survey of autonomic computing—degrees, models, and applications M. C. Huebscher and J. A. McCann
ICL [Aug 2008]
http://portal.acm.org/citation.cfm?id=1380585&dl=GUIDE&coll=GUIDE&CFID=75559502&CFTOKEN=34028179
IBM AC Blue Print:
www-03.ibm.com/.../pdfs/AC%20Blueprint%20White%20Paper%20V7.pdf 
IBM White Paper(s):
//ftp.software.ibm.com/software/tivoli/pdf/itsmstandardsreferencemodel.pdf
ACM: Self-organization algorithms for autonomic systems in the SelfLet approach Davide Devescovi,
Elisabetta Di Nitto, Raffaela Mirandola Univesity of Milano, Dipartimento di Elettronica e Informazione [AUG 2007]
http://portal.acm.
org/citation.cfm?id=1365562.1365597&coll=Portal&dl=GUIDE&CFID=75605482&CFTOKEN=32500398
An Architectural Approach to Autonomic Computing Steve R. White, James E. Hanson, Ian Whalley,
David M. Chess, and Jeffrey O. Kephart IBM Thomas J. Watson Research Center

Autonomic Computing Concepts, Infrastructure & Applications By Manish Parashar & Salim Hariri:
http://books.google.com.pk/books?hl=en&lr=&id=3VtfDzQJUy0C&oi=fnd&pg=PA4&dq=self-learning+algorithms+in+
autonomic+computing&ots=HWm_seld-2&sig=tZAgjKgLS9e3GnAEzztnqErvNsQ#v=onepage&q=&f=false
http://www.research.ibm.com/autonomic/research/projects.html
DARPA SAS:
http://www.darpa.mil/sto/strategic/suosas.html
M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0
M Mo ou uzza
zzam m HHuussa
ssaiinn 2 4 4 3 / FU IE M S / B S C S - 0 7 2443/FUIEMS/BSCS-07
Fe b - 0 4 - 2 0 1 0