How to decide which exception cache size to specify when starting a Detector collection

Products

• Detector for DB2 for z/OS

Issue/Introduction

Specifying Exception Cache size when starting Detector for Db2 for z/OS (PDT) collections.

Resolution

While Detector exception collection is active, contention for the Xmanger address space cross memory lock (CML) may increase the amount of elapsed time in Db2 for SQL statements.
The exception cache pool was introduced to prevent this problem. The exception cache size specified is the number of 1000-byte blocks that are allocated above the line in the Xmanager's private
region. When 0200 is specified, 200,000 bytes are allocated. 0200 is the recommended starting value for a production Db2 subsystem.

Exception cache pool usage can be checked by issuing the following system command:
/F PTXMAN,PSTAT(DTR),DB2(ssid)

When the forementioned system command is issued, the PDT0525 message is written to the Xmanager log.
The PDT0525 message contains an average [cache] hits value. When this value is less than 90%, you should consider increasing
the exception cache size. Most sites will not need to use a value greater than 0500 (500,000 bytes).