Optimize for aggregation
For optimal application performance in FTM it is essential that you make good use of event aggregation. In
most FTM applications all events have the same aggregation threshold which by default is 100. If aggregation
is working at full efficiency then the system will raise the same number of events when processing 1
transaction as it will when processing 100. The easiest way to validate this is to override logging of all
events (set LOG_EVENT_OVERRIDE = 'Y' on the APP_VERSION table record for your application) and run a single
transaction test followed by a 2 transaction test and use the FTM OAC to identify and count all the events
logged for both scenarios. If the number of raised events match for both test cases, then you should repeat
the test for a 100 transaction test case. If the number of raised events do not match, then you must review
the events that were raised from the start to the end of the test case to locate the point in the process that
the aggregation has failed, that is where instead of 1 event referring to 2 transactions you instead see 2
events each pertaining to a single transaction. Once you locate the event that did not aggregate you can
identify the action that raised it as well as the transition and event that triggered the action. You should
then review the model to understand why aggregation is not occurring.
Note: In order to avail of aggregation
for single transaction (non batch) input channels you also need to implement burst mode as described in
Burst mode.
It is also possible to use the FTM level 2 trace to verify aggregation and to
pinpoint points of non aggregation.