Planning CICSPlex SM monitoring

This topic provides some recommendations that will help you plan how to make the best use of CICSPlex® SM’s monitoring functions in your enterprise.

Always have a use for the monitor data you request
Each resource class with a sample interval greater than zero has a CICS® task in each MAS (that is, in each managed CICS system) for which monitoring is required. So, for example, if you activate monitoring of transactions in every CICS system in a CICSplex, each MAS is supporting a CICS task to monitor transactions at the specified interval. Also, CMAS processing is required to maintain the monitor-data cache. You can see that general and unrestricted monitoring would very quickly become an unacceptable overhead. Therefore, do not activate monitoring for any CICS system nor for any CICS resource class without being aware of why you want the data.
Identify resources as specifically as possible
CICSPlex SM monitoring is a specialized function that provides detailed usage and performance data. In general, it will be most useful to you for short-term trend analysis, in anticipating future problems, or dealing with actual problems. For example: However, you should give some thought to understanding which resources are of greatest interest to you. For example, are you really interested in CICS/ESA supplied resources? Often it’s tempting to supply an asterisk (*) as the resource name, which effectively means any instance of the specified resource. If you really aren’t interested in every occurrence, use generic name patterns. For example, you can ask for monitoring of all transactions beginning "PAY" by specifying the resource name "PAY*". Similarly, resources whose status is being reported to the resource status facility should be identified as fully as possible.
Specify as long a sample interval as you can
The shorter the sample interval you specify for a resource class, the greater the cost of gathering the monitoring data. For example, if you set the sample interval to one second for every resource, no other CICSPlex SM work is going to get done at all. Therefore, always specify the longest possible sample interval at which you can obtain useful data. To determine exactly what this interval should be, you must apply what you already know about any particular resource instance. In general, the sample interval should reflect the probable level of use of the resource: if you find that successive sampling of a resource is not showing significant changes, you should increase the sample interval. The sample interval should also have some relationship to the frequency with which the CICSPlex SM or NetView® operator will be looking at the monitor data.
Specify as long a monitor interval as you can
The monitor interval determines the frequency with which counters holding accumulated monitor data are reset to zero. You should set it to a value that is significant in your enterprise, so that the information you get from it is also significant. For example, the interval could be a shift duration, or some other time during which a particular workload is being processed.

You are recommended not to make the interval too short, not only because CICSPlex SM’s overheads grow as the monitor interval reduces, but also because too short an interval is unlikely to provide you with useful statistics.

Do not define more monitor specifications than you really need
Avoid having too many monitor specifications, not only because of the extra work involved in creating them and associating them with CICS systems, but also because a CICS system can belong to only one monitor specification at a time. The more monitor specifications you define, the smaller the opportunity for establishing a general-purpose monitoring policy.
Specify a suitable retention period
You don’t have to specify a retention period. However, the retention period defaults to zero minutes, and so as soon as monitoring stops for any CICS system, that system’s monitor data is lost. If some unexpected event occurs, the data could be invaluable, so setting the retention period greater than zero is advisable. Note, however, that if a CICS system stops, monitor data is lost when you restart the system regardless of any retention period you may have specified. In this case, it’s important to look at the monitor data before you restart the CICS system.

You can set the retention period to any value between 1 and 1440 minutes (24 hours). You can specify a retention period on the monitor specification, in which case that period applies to every CICS system associated with the monitor specification. Alternatively, you can specify a retention period in the CICSPlex SM CICS system definition.

Decide when to activate monitor definitions
When you add a monitor definition to a monitor group, you can specify the part of the day during which it is to be active. You should make considered use of this feature. For example, you probably will not want monitoring active overnight, or during any period when general use is low and no one is around to view the data, so ensure that monitor definitions are not in effect at those times. Also, if you’re monitoring resources for which there is contention, think about when this contention occurs. Is it constant, or does it occur only between 0900 and 1100 hours? If the latter, limit your monitoring to that period.
[[ Contents Previous Page | Next Page Index ]]