Reporting Examples › 6.2 The Ten Largest Consumers of CICS Service Units Last Month

6.2 The Ten Largest Consumers of CICS Service Units Last Month

Consider the CICS information area.  Within this information
area, look for a file that distinguishes service units by
user, which we know is in the CICS account variable CICACT3.
A review of the CIC information area files (in the CA MICS
Analyzer Option for CICS Guide) shows that the CICS User
Activity file (CICCSU) and the CICS Application Unit Activity
file (CICCAU) are likely candidates.  Both contain fffSERVU
data elements where fff is either CSU or CAU.

Since we are interested in a month, we need the MONTHS
timespan.  Since only the CICCSU file is active in the MONTHS
timespan (by default), we must use this file.  Since we want
last month, we use cycle 01.

We decide to print CICACT3 and CSUSERVU.

The SAS program and output are as follows:

  PROC SORT DATA=&PCICM..CICCSU01 OUT=CICMAX;
      BY DESCENDING CSUSERVU;
  DATA;
  SET CICMAX;
  IF_N_<11;
  PROC PRINT;
      VAR CSUACT3 CSUSERVU;
  RUN;


  OBS     CICACT3      CSUSERVU

   1       MFC304      14303667
   2       XOT704       8958395
   3       MPS526       3720106
   4       EOT419       2915726
   5       FVB001       2077608
   6       MPS123       2014002
   7       CVC002       1417266
   8       MPS523       1136576
   9       LSG005        987786
  10       CVC004        969125


Important!  Note that the program begins with a sort and uses
the option DESCENDING, which sorts CSUSERVU in descending
order.  If we had omitted the OUT= option, we would have
replaced the MONTHS timespan of the CICCSU01 data set with
the result of the sort.  This would have caused a serious CA
MICS database integrity problem.  PROC SORT should never be
used on a CA MICS data set without an OUT= option.  This
eliminates any possibility of affecting the integrity of the
database.

The variable _N_ is a specially defined SAS variable that
counts the number of observations processed in a DATA step.
Since we have already sorted on CSUSERVU, we want the first
ten observations only for later printing with PROC PRINT.

This program and resulting output look simple and
straight-forward, but, unfortunately, we may not get the
answer we need.

Look at the CICCSU file organization in the Files chapter in
the Analyzer Option for CICS Guide. This shows the sort order
and summarization level for the CICCSU file in the MONTHS
timespan.   Note that ZONE is last in the sort order and is
THE LOWEST LEVEL OF SUMMARIZATION.

Assume that there are three zones in the file.  Our SAS
program ignored ZONE.  That is, assuming a user signed on in
more than one ZONE, he had more than one observation in the
file.  Therefore, it is possible that we did not consider the
total CICS service units consumed for every user. Most use of
CICS would be in prime time (ZONE='1'), but there could be
some people (systems programmers for example) who use
significant CICS resources on other shifts and weekends. It
is the total CICS service units consumed by these people that
we want to report.  If their total use were high enough, they
should appear in the printed output. Therefore considering
their use over all zones might change the result.

The question our program answered was, "Who were the ten
highest users of CICS service units IN ANY ZONE ON ANY SYSTEM
last month?"  If this is satisfactory for our purposes, we
are finished.  However, if we want to answer the original
question, we have not yet done so.  If there were more than
one SYSID, we would also have a similar problem.

We could answer the question by using the CA MICS
summarization facility to summarize ZONE to MONTH (and if
necessary SYSTEM).  Then we could use our original program.

More Information:

Using Summarization