You can specify single statistics, multiple statistics,
or groups of related statistics for monitoring through the Performance
Monitoring Infrastructure by using the administrative console.
Before you begin
Ensure that you have enabled performance monitoring, and that
you have at least once invoked the event you want to monitor before
performing this task.
Procedure
- Open the administrative console.
- Select .
- Select the server or node agent that contains the event
points that you want to monitor.
Note: You cannot choose
to monitor statistics on a cluster; you can only do so on a specific
server or node.
- Expand some of the groups, such as WBIStats.RootGroup or Enterprise
Beans. All of the statistics that can be monitored
are within the listed groups. Some statistics may not be listed because
they have not been invoked since the server was last started.
- Select a statistic you want to monitor from within the
tree on the left side of the panel, and then select the statistics
that you want to collect on the right side, then click Enable.
Repeat this for all statistics that you want to monitor.
- Go back to the server PMI configuration page by clicking
the server name link.
- Click Apply or OK.
- Click Save.
Results
You can now start monitoring the performance of your chosen
statistics in the Tivoli® Performance
Viewer.
Note: When viewing these statistics,
you should not mix counter-type statistics with duration-type statistics.
Counters are cumulative, and the scales against which they are graphed
them can quickly grow depending on your application. Duration statistics,
in contrast, tend to remain within a certain range because they represent
the average amount of time that it takes your system to process each
event. Consequently, the disparity between the statistics and their
relative scales may cause one or the other type of statistic to appear
skewed in the viewer graph.