EventMonitor Properties

Multiple engine property values may be changed to alter the performance of the EventMonitor service, including:
  • EVENT_MONITOR_EVENT_LIMIT_PER_RUN
  • EVENT_MONITOR_EVENT_COUNT_FOR_CATCHUP
  • EVENT_MONITOR_THREADS
  • EVENT_MONITOR_JUST_SLC

    Available from v6.1.2.1 and above

EVENT_MONITOR_EVENT_LIMIT_PER_RUN and EVENT_MONITOR_EVENT_COUNT_FOR_CATCHUP dictate how much data the EventMonitor service will attempt to read from the EVENTS table in one batch, and the count of records retrieved at which it will enter 'catchup mode', respectively. Note that in 'catchup mode' the EventMonitor service will not wait before querying for more data to process.

As we learn more the default values for these parameters are changing. We currently recommend setting EVENT_MONITOR_LIMIT_PER_RUN to 1000 and setting EVENT_MONITOR_EVENT_COUNT_FOR_CATCHUP to 200 for best performance.

TIP

The EVENT_MONITOR_COUNT_FOR_CATCHUP value should always be made smaller than the value for EVENT_MONITOR_EVENT_LIMIT_PER_RUN for best performance. This way even if the EventMonitor does not get limit events to process, if it gets a big percentage of the limit it will immediately look for more to process rather than waiting before it looks again.

EVENT_MONITOR_THREADS, available from v6.1 and above, whose default is the number of CPUs the server running the EP has. Default value for EVENT_MONITOR_THREADS was 2. This property should reduce the amount of time spent constructing XMLStrings as its value is increased. However, depending on the number of CPUs allocated to the server running the IBM® Sterling Control Center Monitor CEP it may, or may not, help overall EventMonitor performance to increase its value.

EVENT_MONITOR_JUST_SLC, available from v6.1.2.1, may be set true when no Connect:Direct File Agents are monitored by IBM Sterling Control Center Monitor. Setting this property to true allows the EventMonitor logic to limit the event data retrieved, and therefore the events created and passed on, which allows it to run faster and decrease the delta between ACTION_COMPLETED and reconstituting events.

TIP

If SLCs are incorrectly generating events about things not happening, check the maximum and average milliseconds delay metric, added to v6.2.1. If either value is large you may need to:
  • Ensure IBM Sterling Control Center Monitor database table indices are being maintained
  • Increase EVENT_MONITOR_LIMIT_PER_RUN and EVENT_MONITOR_EVENT_COUNT_FOR_CATCHUP
  • Increase EVENT_MONITOR_THREADS
  • Set EVENT_MONITOR_JUST_SLC if no Connect:Direct File Agents are monitored