Performance monitoring of processes

The trigger for starting Process Performance monitoring is based on events that occur for the ProcessorGroup and MemoryGroup instances. (Chapter 11, “Environmental Objects and Host Resource Monitoring,” provides more information about ProcessGroup and MemoryGroup topology.)

The Server Manager computes the performance monitoring of processes for unique application instances on a given Host.

• If there is a single instance of a process or an application on the Host, Server Manager computes the performance monitoring for it.

• If there are multiple instances of a process or an application on the Host, Server Manager does not update the CPU/Memory utilization values. Instead, the multiple instances of the process are treated as a single Application.

  To verify the current count of processes, right-click on the Host instance and select Configure Process Monitoring in the Domain Manager Administration Console.

  Metrics are collected:

• CPU metrics —When the HighUtilization event for the ProcessorGroup instance is active.

• Memory metrics —When the InsufficientFreeMemory event for any MemoryGroup instance is active.

  Process CPU/Memory metric collection stops when the related event on the Host ProcessorGroup/MemoryGroup becomes inactive or when the application becomes unavailable.

  The following attributes on the Host instance in ESM reflect whether these events are in active state:

• IsProcessorGroupUtilizationHigh —Set to TRUE whenever the ProcessorGroup::HighUtilization event is active.

• IsMemoryGroupUtilizationHigh —Set to TRUE whenever the MemoryGroup::InsufficientFreeMemory event is active.

  Note:

  The computation of these attributes also considers the Smoothing Intervals that are set for the events.

  The polling for CPU metrics stops if the value of IsProcessorGroupUtilizationHigh becomes FALSE. If CPU metrics were being polled for a given system, the CPUUtilization values on Application instances will be reset.

  The polling for Memory metrics stop if the value of IsMemoryGroupUtilizationHigh becomes FALSE. If Memory metrics were being polled for a given system, the MemoryUtilization values on Application instances will be reset.

  For devices discovered via SNMP, the values of IsProcessorGroupUtilizationHigh and IsMemoryGroupUtilizationHigh attributes are updated depending on the source of the event for the ProcessorGroup and MemoryGroup classes.

  If the IP Domain Manager is deployed as an IP Availability and Performance Manager and the discovery and monitoring of resource entities are disabled in Server Manager:

• The Server Manager subscribes to events on ProcessorGroup and MemoryGroup from the IP Domain Manager.

• The IsProcessorGroupUtilizationHigh and IsMemoryGroupUtilizationHigh attributes are updated based on their event states in the IP Domain Manager.

  In this case, you should configure the thresholds for these entities in the IP Domain Manager.

  If the IP Domain Manager is deployed as an IP Availability Manager and the discovery and monitoring of resource entities are enabled in Server Manager:

• The Server Manager looks at events on ProcessorGroup and MemoryGroup locally in the Server Manager.

• The IsProcessorGroupUtilizationHigh and IsMemoryGroupUtilizationHigh attributes are updated based on their event states in the Server Manager.

  In this case, you should configure of the thresholds for these entities in the Server Manager.

  Having the discovery and monitoring of resource entities turned on in both Server Manager and IP Domain Manager will result in duplicate polling, duplicate notifications, and improper computation of root-cause events that are part of process performance monitoring.

  Configuring Host Resource monitoring on page 261 provides information about configuring resource discovery in the Server Manager.