.IMS

.IMS contains all of the modules the IMS subsystem uses. This includes IMS Full Function (DFSxxxxx), IMS Fast Path (DBFxxxxx), and DBRC (DSPxxxxx) modules. If the IRLM (DXRxxxxx) is being used for local and/or global locking, these modules will be shown in section .DB2IRLM.

It is not unusual or even undesirable for IMS modules to use 25% to 90% of the CPU for the entire job. If the application code is well tuned, the database calling patterns are well tuned, and the database buffer pool is well tuned, most of the CPU use should be in .IMS.

If IMS uses less than 25% of the CPU, especially in a BMP or batch environment, this might indicate that the CPU is being used inefficiently elsewhere in the job.

Hints

In an IMS environment, performance improvement should be performed concurrently with standard IMS performance tuning. iStrobe can help to indicate when IMS performance improvement is possible.

In an IMS batch environment the Resource Demand Distribution report will show activity for the IMS log (IEFRDER and IEFRDER2) and the RECON data sets.

The Wait Time by Module report can show wait in IMS modules or in SVCs that IMS is using.

The Data Set Characteristics report will show I/Os for OSAM and VSAM databases, and VSAM buffers, which can indicate the need for performance improvement in a database buffer pool and call pattern. It will also show I/Os for STEPLIB data sets. Performance analysis of STEPLIB in an IMS Message Region is an important activity for improving IMS performance.

The I/O Facility Utilization report can show contention as a result of too many highly used databases and data sets on the same DASD volume.

The I/O Use by Resource report can show concentrations of database I/O activity down to a cylinder level.

If the Program CPU Use report shows that CPU is being used in the IMS call action modules, investigate database calling patterns and database buffering. This is discussed in the descriptions of the individual modules. In an IMS batch environment, use the IMS DBMONITOR and the database call trace (DLITRACE LOG=YES,COMP in the DFSVSAMP control cards). In a BMP or MPP environment, use the IMS DCMONITOR and the database call traces (/TRACE SET ON PSB xxxxxxxx COMP for database calls and /TRACE SET ON PGM xxxxxxxx for data communications calls). The IMSASAP II TRACE reports from the IMS DCMONITOR are useful.

All these examples show how iStrobe can identify areas of IMS performance analysis that you should investigate to provide the greatest improvements in performance.

If an IMS module is using an excessive amount of CPU time, or you are curious about what a particular IMS module is doing, analyze the CPU use by recreating the Performance Profile with the module of interest detailed. You can then compile the IMS modules and match the offsets to the areas of CPU use.

Reference Sources

Strobe Interpretation and Analysis User Guide, Compuware Corporation.