IMS internal resource usage
There are several summary reports that you can use to examine the level of internal contention for resources.
The following list gives a brief explanation of these reports.
- Pool Space Failure Summary report
- The Pool Space Failure Summary report gives the number of times
in each region a given amount of storage was unavailable. It shows
the number of bytes and the identification of the pool as well as
the number of occurrences of this failure to obtain storage. You can
use this summary to determine whether you need to increase the buffer
pool allocation by a system definition change or by overriding the
number of buffers in the EXEC statements in the JCL.
The format of the report is shown in the following example.
POOL SPACE FAILURE SUMMARY POOL ID BYTES REQ. OCCURRENCES DLMP 8888 1 TOTAL 1 - Latch Conflict Statistics report
- The basic serialization of the task processing in IMS is controlled by ownership of an IMS latch. When different programs are executing,
they compete for the ownership. If they wait for the resource, the
one possessing the latch has to post the other ITASK waiting for it.
You can judge the level of contention for a resource and then investigate
a set of changes to relieve the pressure.
The different types of latches and the counters that exhibit the level of contention are given in the Latch Conflict Statistics report. The following figure is an example of this report. The entries are organized according to the latch names.
When a system checkpoint is taken during the time the monitor is active, latch conflict statistics are reset to zero, thus corrupting the values presented in this report. If this situation exists, the following message will be inserted at the top of the report:**** A CHECKPOINT OCCURRED DURING MONITOR RUN **** **** LATCH CONFLICT STATISTICS ARE INVALID **** **** SEE UTILITIES REFERENCE MANUAL ****However, if the master terminal operator issues the /CHECKPOINT command with the STATISTICS keyword parameter, latch conflict statistics are reset to zero, but the IMS monitor is not notified. Therefore, DFSUTR20 cannot detect that the statistics have been corrupted and does not issue this message.
Recommendation: Do not issue statistics checkpoints while the Monitor is running.The counters are primarily concerned with storage management and logging services. The statistics recorded are the number of times contentions occur, that is, the resource waits for a latch.
IMS MONITOR ** GENERAL REPORTS ** TRACE START 1993 209... LATCH CONFLICT STATISTICS LATCH COUNT AT AT NAMES FIELD START END DIFF. LOGL CONTENTIONS 0 0 0 SMGT CONTENTIONS 0 0 0 XCNQ CONTENTIONS 0 0 0 ACTL CONTENTIONS 0 0 0 CBTS CONTENTIONS 0 0 0 DBLK CONTENTIONS 0 0 0