We've discussed at length DASD monitoring considerations, and the cost of monitoring when it comes to DASD performance metric collection. What I would like to do next is take a look at other aspects of the cost of monitoring, and take into account such things as optimization of CPU required for monitoring.
One of the first things to look at when it comes to looking at the cost of monitoring is if you and your users are employing Autorefresh. Autorefresh implies that OMEGAMON will be regenerating a given monitoring display (screenspace, CUA screen, or TEP workspace) on a timed interval. In Classic and CUA interface, Autorefresh is set in the session options, and if used extensively, Autorefresh can measurably drive up the the CPU usage of the Classic and CUA tasks. For example, if you have multiple users, each running a display on a relatively tight (10 seconds or less) interval, OMEGAMON is doing a lot of work just painting and re-painting screens on a continuous basis.
The recommendations are as follows:
Limit the use of Autorefresh
If you must use Autorefresh, set it on a higher interval (60 seconds or higher)
Better yet, if you must use Autorefresh, us the Tivoli Portal to drive the display. The TEP has more options to control Autorefresh, and you will be moving some of the cost of screen rendering from the collection tasks to the TEP infrastructure.