SAS Programming

Guidelines Interview
Questions
Review these typical interview questions and think about how you would
answer them. Read the answers listed; you will find best possible answers
along with strategies and suggestions.
This page is intentionally left blank
 Chapter 1

Efficient SAS Programming

1: How do you achieve scalability in SAS programming?

Answer:
SAS program scalability can be achieved in 2 ways - by scaling up
and scaling out. Scalability is ensuring the lowest time to solution,
especially for the most vital tasks. Typically, when you want to
speed up the task completion, you either try to complete multiple
processes at the same time or distribute the task across various
processors and do parallel processing. This, sometimes, involve
overlapping of certain processes. Scaling up requires better
hardware that is capable of multiprocessing which is known as
symmetric multiprocessing or SMP. Scaling out requires more
servers that can handle distributed processing.
2: How do SQL Views help better efficiency?
Answer:
A View typically consists of a subset of the entire table and hence
is more efficient as it accesses a smaller set of data which is
required. View also lets you hide the sensitive columns and
complex queries from the user by choosing only what needs to be
shown. Views always fetch fresh data from the table as they do
not store any data.

3: What do you know about the SPD Engine?

Answer:
The SPD Engine or the SAS Scalable Performance Data Engine is
developed for SAS 9 to speed up the processing of large data sets
by splitting them into smaller physical files called partitions.
There are several parallel processors that have exclusive access to
each partition and process them in parallel using threads.
Partitions are created when the SAS data sets are created. When a
Where clause is mentioned it is split across the partitions and
processed in parallel. Data blocks are also read in parallel.
Multiple connections are created based on the partitions which
further reduces the I/O bottlenecks. The SPD Engine also does an
implicit sort if the query contains a by clause.

4: What resources are used to run a SAS program?

Answer:
The six resources used to run a SAS program are:
a) Programmer time: The amount of time taken by the
programmer for writing, testing and maintaining the
program
 b) Real time: The time elapsed while executing a job
c) CPU time: The amount of time the CPU takes to perform a
task. The task can be reading data, writing data, calculations
or implementation of a logic
d) Memory: The work area memory space used for holding
executable programs, data, etc
e) Data storage space: The disk space for storing the data. This
is measured in terms of bytes, kilobytes, gigabytes etc.
f) I/O: The read and write operations performed to movie data
from the memory to any output device, and vice versa

5: List the factors that need to be considered while assessing the

technical environment.
Answer:
The four factors that need to be considered while assessing a
technical environment are:
a) Hardware: Available memory, number of CPU’s, number of
devices connected, network bandwidth, I/O bandwidth, and
capability to upgrade
b) Operating environment: The resource allocation & I/O
methods
c) System load: This includes the number of users sharing the
system, the network traffic, and the predicted increase in
load
d) SAS environment: includes all SAS software products
installed, number of CPU’s, and memory allocated for SAS
programming
6: Explain the functionality of the system option STIMER in the
Windows environment.
Answer:
STIMER option in the Windows environment specifies that CPU
time and real time statistics are tracked and written to the SAS log
throughout the SAS session.
Example: The following line of code turns on the STIMER option.
options STIMER;

7: What is the function of the option FULLSTIMER in the

Windows operating environment?
Answer:
FULLSTIMER option in the Windows environment specifies that
all the available resource usage statistics needs to be tracked and
written to SAS log throughout the SAS session.
Example:
options FULLSTIMER;

8: Explain the MEMRPT option.

Answer:
The MEMRPT option in the z/OS environment specifies that the
memory usage statistics are tracked and written to SAS log
throughout the SAS session. This is not available as a separate
option in the Windows operating environment.
9: While benchmarking the programming techniques in SAS,
why is it necessary to execute each programming technique in
separate sessions?
Answer:
It is always necessary to execute each programming technique in
separate SAS sessions while benchmarking them the first time a
program is read because the operating system might load the code
into the cache and retrieve it from the cache when it is referenced.
This takes less time. The resource usage necessary to perform this
action is referred to as overhead. Using separate sessions minimize
the effect of overhead on resource statistics.

10: While doing benchmark tests, when is it advisable to run the

code for each programming technique several times?
Answer:
It is advised to run the code for each programming technique
several times while benchmarking tests if the system is executing
other jobs at the same time. Running the code several times
reduces variations in the resource consumption associated with
the task and so the average resource usage is known.

11: How do you turn off the FULLSTIMER option?

Answer:
The FULLSTIMER option can be turned off with the following line
of code.
options nofullstimer;
12: What steps can be taken to reduce the programmer time?
Answer:
Programmer time is the amount of time required for the
programmer to determine the specifications, write, submit, test
and maintain the program. It is difficult to calculate the exact time,
but it can be reduced by the use of well-documented
programming practices and reuse of SAS code modules.
 Chapter 2

Memory Usage

13: What is PDV? How does it work?

Answer:
PDV or Program Data Vector is a memory area created after the
input buffer is created. Two extra variables _N_ and _Error_ are
created by the SAS engine during compilation. These variables are
used for processing but never written into the data set. SAS
creates a PDV for each observation.

14: How would you choose between Data Step and Proc SQL?
Answer:
With small data sets, Proc SQL works better since it loads the
entire data set into the memory and works with the data. So
there’s less need to go back and forth into the database. But with
large data sets Data Step will work better as loading the entire
data set with Proc SQL will block a huge chunk of memory. Data
Step will always take one record at a time and hence, the number
of records or large volume of data will not matter as long as the
database connectivity remains good.

15: Explain memory management in SAS.

Answer:
SAS, unlike Java and .Net, does not have garbage collection for
memory management. But it does accomplish the job with a series
of instructions called steps. Memory is allocated when the step
begins and released when the step completes. This way, there’s no
memory loosely allocated during the runtime. When dealing with
large volumes of data, there may be cases when ample memory is
not available. In such cases, SAS pushes an error message that
memory not available, which is logged for reference. The hash
objects in SAS lets you handle considerable amount of objects
quickly. The Data Step is also efficient in memory management as
it takes only one record at a time. Since most of the SAS programs
depend upon a Work Area which they use to store objects
temporarily, this area typically runs out of memory which needs
to be handled efficiently.

16: What is the sequence of action performed in the background

while trying to create a data set from another data set?
Answer:
While creating a data set from another data set the following
actions take place in the background
a) The data gets copied from the input data set to a buffer in
memory
b) From the input buffer an observation at a time is written to
PDV (Program Data Vector)
 c) Each observation from PDV is written to output buffer when
processing is complete
d) The contents of output buffer are written to disk when the
buffer is full.

17: Define PAGE and PAGESIZE.

Answer:
A PAGE is a unit that indicates the data transfer between a storage
device and PAGESIZE is the amount of data that can be
transferred to one buffer in a single I/O operation.

18: What procedure is used to indicate the PAGESIZE of a data

set?
Answer:
The Contents procedure is used to know the PAGESIZE associated
with a data set.
Example: The following Contents procedure issues a message to
SAS log indicating the PAGESIZE associated with the data set
exam.clinic1. This also gives the number of data set pages.
Proc contents data = exam.clinic1;
run;

19: Is it possible to control the PAGESIZE of an output data set?

Answer:
It is possible to control the PAGESIZE of an output data set by
using BUFSIZE= option, which specifies the PAGESIZE in bytes
Example: The following program creates a data set exam.clinic1
from the data set exam.clinic2. In the following program the
BUFSIZE= option specifies a PAGESIZE of 30720 bytes.
options bufsize=30720;
libname exam ‘c:\myprog’;
data exam.clinic1
set exam.clinic2;
run;

20: What is the default value of the BUFSIZE= option?

Answer:
The default value of the BUFSIZE= option is 0. If BUFSIZE= option
is set to zero SAS uses the optimal PAGESIZE determined by SAS
for that operating environment.

21: Is it necessary to specify the BUFSIZE= option every time a

data set is processed?
Answer:
No. The BUFSIZE= option is set at the time of creation of data set,
and that value of becomes a permanent attribute of the data set.
Once it is specified it is used every time the data set is processed.

22: What does the BUFNO= option signify?

Answer:
The BUFNO = option is used along with an SAS data set to lay
down how many buffers are available for reading, writing, or
updating. The larger the value of BUFNO = the faster the
input/output function would be since more values will be stored
in the buffer which avoids an actual input/output function. You
can specify larger number of pages to include in the BUFNO = and
accordingly that many pages will be loaded into the memory.
For Example, the following program creates a data set
MyExam.MyClinic from the data set MyExam.MyClinic2 in the
following program, the BUFNO = option is given the value 6, that
denotes 6 buffers.
options bufno=6;
libname exam ‘D:\MyProgram’;
data MyExam.MyClinic
set MyExam.MyClinic2;
run;

23: How do you set the BUFNO= option to the maximum

possible number?
Answer:
To set the maximum value to BUFNO= option, you can set
BUFNO= MAX which sets the maximum buffer value available in
the current operating environment. The largest possible value of
max would be approximately 2 billion (231-1).
For Example, the following program creates a data set
MyExam.MyClinic1 from the data set MyExam.MyClinic2. In the
following program, the BUFNO = option is given the value max,
that denotes the maximum buffer available in the current
environment.
options bufno=max;
libname exam ‘D:\MyProgram’;
data MyExam.MyClinic
set MyExam.MyClinic2;
run;
24: Is it necessary to specify the BUFNO= option every time a
data set is processed?
Answer:
It is mandatory to specify the BUFNO = option every time a data
set is processed. This is required since the buffer varies every time
a data set is opened and closed. Moreover, the BUFNO = value set
is valid only while a data set is open in the current session.

25: What are the general guidelines for specifying the buffer size
and buffer number in the case of small data sets?
Answer:
The main objective behind specifying the buffer size and buffer
number is to reduce the number of I/O operations. In the case of
small data sets, care must always be taken to allocate as many
buffers as there are pages in the data set. This ensures that the
entire data set can be loaded into the memory using a single I/O
operation.

26: How does the BUFSIZE= and BUFNO= impact the following
program?
data exam.clinic1 (BUFSIZE=12288 BUFNO=10);
set exam.clinic2;
run;
Answer:
The above program reads the data set exam.clinic2 and creates
exam.clinic1. The BUFSIZE= option specifies that exam.clinic1 is
created with a buffer size of 12288 bytes. The BUFNO= option
specifies that 10 pages of data are loaded into memory with each
I/O transfer.
27: Explain the SASFILE statement.
Answer:
The SASFILE statement loads the SAS data file into the memory to
be available further to the program. With SAS File you can free the
buffers. Instead, the file is loaded and kept in the system memory
with a pointer in the program to access it.
The following example explains the use of SASFILE in a simple
way. The SASFILE statement opens the data set MyExam.MyClinic
and allocates the buffer. It reads the file and loads it into the
memory so that it is available to both the PROC print as well as the
PROC MEANS step. Finally, the SASFILE data file is closed with
the close statement and the buffer is cleared.
SASFILE MyExam.MyClinic load;
proc print data= MyExam.MyClinic
var. Serial No result;
run;
proc means data= MyExam.MyClinic;
run;
SASFILE MyExam.MyClinic close;

28: What happens if the size of the file in the memory increases
during the execution of SASFILE statement?
Answer:
When the SASFILE statement is executed, SAS assigns some buffer
to the DATAFILE based on the number of pages to be loaded and
the size of the index file. Once this is done, the file data is loaded
into the memory for updates. The buffer size is automatically
increased as the file size to be saved increases. The initial buffer
memory size allocated is only the minimum memory allocated to
load the file. It automatically increases provided there’s ample
memory left in the current operating system.

29: Mention the guidelines to be followed while using SASFILE

statement.
Answer:
While using the SASFILE statement, the following procedures are
to be followed:
a) There should be sufficient real memory to load the file
b) In case, there’s not enough memory to load the entire file
into one SAS data set, the data step should be used to create
a subset of the file which will fit into the available memory.
Since one part of the file is already loaded into the memory,
the rest of the file data can also be easily accessed by the
program. This reduces the CPU time significantly.

30: When is the buffer allocated by the SASFILE statement

freed?
Answer:
The buffer allocated by the SASFILE statement to load the data file
is freed in two instances:
a) When the SASFILE Close statement is executed, the file is
closed, and the buffer allocated for the DATAFILE is closed.
For example, In the following program the SASFILE
statement opens the data set. MyExam.MyClinic and
allocates the buffer. It reads the data into the memory which
is available to the PROC print and PROC MEANS steps. The
last SASFILE statement closes the SAS DATAFILE and frees
the buffer allocated for the file.
SASFILE MyExam.MyClinic load;
 proc print data= MyExam.MyClinic
var Serial No result;
run;
proc means data= MyExam.MyClinic;
run;
SASFILE MyExam.MyClinic close;
b) The SASFILE buffer is allocated only as long as the session is
open. When SAS session ends, it frees the buffer and closes
the DATAFILE.

31: Which operations are not allowed in a file opened with

SASFILE statement?
Answer:
There are certain operations that cannot be performed on a file
opened with SASFILE statement, such as replacing the file and
renaming the variables.

32: How do you calculate the total number of bytes occupied by

a data file if you know the PAGESIZE?
Answer:
The total number of bytes that a data file occupies can be
calculated by multiplying the PAGESIZE by the number of pages.
Example: If the data file exam.clinic1 has a PAGESIZE of 8192 and
number of pages is 900, then the data file occupies 7372800 bytes
(8192* 9423).
This page is intentionally left blank