1. Set object auditing to *NONE for all Collector software objects:

CHGOBJAUD  OBJ(LSYSLOG/*ALL)  OBJTYPE(*ALL) OBJAUD(*NONE) 

This is the most likely cause of performance problems. Just running the collector is generating journal entries which is keeping the monitor and transformer jobs busy.  On the development system (a low end iSeries with very little activity other than the collector running), you can have more than 64,000 ZC journal entries in 20 minutes.

2. If not already used, add *NOQTEMP setting to system value QAUDCTL.

3. Set Verbose Logging to No for the Collector software (Destination Configuration).

4. Change the number of audit transformers to at least 5.  Monitor activity and adjust from there. (Monitor Configuration)

5. Use only the start/stop/configure commands supplied by TIBCO Software to start and end the Collector jobs.  When the destination and monitors are started/stopped, control information is logged in files so that restart will not skip any logs. Results are unpredictable if component jobs are started/stopped with system work management commands.

6. Reconsider the use of the system value QCRTOBJAUD(*CHANGE).  This turns on object auditing on any object created on the system (where the library CRTOBJAUD value is *SYSVAL). 

7. Reconsider the use of object auditing in general. See the IBM Security Reference (Chapter 9; and Info Center) for recommendations. Auditing changes to every object on the system generates many, many audit journal entries and can potentially affect performance.