Previous Topic: How Primary Data Collection Is Synchronized in a ClusterNext Topic: Support Real-time Performance Agents


Alternate Performance Agents

You can create alternate Performance Agents to collect specific sets of data for special purposes. For example, you can specify an interval of 10 to 15 seconds with only system metrics data to be collected. You could run this Performance Agent for short periods of time during unusually heavy system demands to analyze these demands.

You can also create a Performance Agent to record data based on a longer interval such as 1800 to 3600 seconds. This Performance Agent would produce smaller data files that would act as an alternative to archiving large data files. See the note following this section.

The primary Performance Agent now acts as a data server to all other data collection processes. You can define additional collection processes to collect subsets of data from classes at different intervals and record this data for performance analysis or for dynamic viewing:

Note: The primary data collection process must be running for any of the other Performance Agents to have access to current data. The primary Performance Agent doesn't have to be actually collecting data to act as a data server (the data collection process may be in a waiting state due to a scheduled wait).

If you start an alternate Performance Agent or Real-time Performance Agent while the primary Performance Agent is not running, each Performance Agent writes a message in the PSDC$DC.LOG file and exits. This will also happen if the primary Performance Agent is stopped intentionally while other Performance Agents are running.