Comparing Session Results

After making changes to your CICS region and its transaction-processing programs, conduct a measurement session to confirm that the changes have produced the necessary results.

If you conduct your measurement sessions in a production environment, the circumstances of the two sessions will not be exactly the same. The factors that you can control to make the sessions more comparable include: duration of the session, the time of day, the day of the session, and the number of samples collected. Other factors that you cannot control but affect performance include: stretch time, paging rates, the volume of transactions, and the transaction mix. By making the circumstances as similar as possible, you can more accurately assess whether your changes improved application performance.

The relative CPU usages shown in the reports that deal with CPU usage percentages alone will change, but not necessarily in the direction you might expect. Because CPU usage percentages must total 100%, a large reduction in the percentage of CPU usage in one module must be accompanied by increases in percentages of CPU usage in other modules, though the latter’s absolute CPU usages remain unchanged.

If your two sessions are comparable, then you should see a reduction in the absolute measures of resource usage that you have tried to affect—CPU usage and EXCP counts or both—in the Measurement Session Data report. If you have reduced CPU usage, the percentage of wait time will have increased correspondingly and response times should improve and become more consistent.