I’m research’g the config of Entirex due to CPU usage increase since we upgraded. CPU seconds doubled. We are using RPC and are running as a batch job. Should Broker run as a started task rather than a batch job? Since we are running RPC, should Broker be placed into it’s own performance class or group? Is the increase due to contention with other batch jobs running in the same region since CPU usage decreases on weekends with the same number of transactions (when Broker is one of the only jobs running)? What parameters (z/OS or Broker) might I tweak to get better performance and less CPU usage?
what did you upgrade from/to?
Broker is normally run as a started task. Its performance group is typically in the same range as Adabas and CICS, since it is servicing online, interactive users. Batch RPC Servers (EntireX or Natural) are usually in a lower performance group, but still above general batch.
We went from 7.2.1.34 => 7.3.2.004. The engineers did an intermediate upgrade (7.2.1.50) prior to applying the 7.3 version all in one night.
There are some versions since 7.3.2 (7.3.4 and 8.0), so I’d try upgrading to the latest and retest. If there is still a significant rise in CPU usage, a support request would be in order.
You should compare the old 7.2 attribute file with the new 7.3 attribute file. Maybe some parameter has changed due to the migration step (e.g. translation/conversion).