Implementing RTA

Real-time analysis should be implemented gradually and in response to an established need for the resultant information, especially if you are not planning to automate your responses to exception conditions. You are recommended to begin by implementing SAM, and to implement MRM after having analyzed records, such as problem logs, that give you some indication of where MRM would be useful.

Implementing system availability monitoring (SAM)

You are recommended to implement SAM as follows:

  1. Create time-period definitions (PERIODEFs) for the expected availability times of each CICS® system to be managed by CICSPlex® SM. For example, if seven of your CICS systems should be active from 0800 hours through 1900 hours, create one time-period definition for that part of the day.
  2. When you define each CICS system to CICSPlex SM, supply the name of the time-period definition that defines the system’s hours of availability.

These two actions are sufficient to activate SAM and to cause the default external notifications to be generated when a noteworthy condition occurs. As you can see, you can implement SAM as soon as CICSPlex SM is installed in your enterprise, and with very little additional effort. You are recommended to continue using SAM with the default notifications, at least until you have had an opportunity to see the default notifications of all SAM conditions. If you decide to customize any of the notifications, start by creating an action definition for one of the SAM conditions and apply it to a single CICS system to test the effects. When you are happy with the change, apply it to other SAM conditions and other CICS systems, as appropriate.

The overheads associated with SAM are not high, so you can activate it throughout the enterprise CICSplexes without affecting the general performance of CICSPlex SM or the CICS systems it manages.

Implementing MAS resource monitoring (MRM)

You should start by implementing MRM on a system-by-system basis. Select a single resource type, such as connections or files, in a single CICS system, and create these CICSPlex SM objects:

The interval between resource evaluations (as specified in the evaluation definition) and the intervals at which data is to be analyzed (as specified in the analysis definition) should be intelligently matched (and ideally the same). In particular, do not analyze the data more frequently than it is evaluated. Also, resource evaluations should be performed as infrequently as possible while still providing useful data: the more frequently data is collected, the greater the cost of its collection.

Begin by installing the analysis definition manually in the active CICS system. When you are satisfied that MAS resource monitoring is working as you expected, consider whether automatic installation of the analysis definition would be useful. If it would, add the analysis definition to an analysis group, and associate the analysis group with an analysis specification. At this point, decide whether to monitor additional resources in the same CICS system. To extend MAS resource monitoring to other resources, an additional evaluation definition would be required, an additional analysis definition might be required, but the original action definition could probably be reused.

Implementing analysis point monitoring (APM)

When you are happy with the MAS resource monitoring results, progress to analysis point monitoring (APM), which is applied to a logical grouping of CICS systems. You should consider which CICSplexes are affected by your analysis point monitoring requirements, and identify the CMASs involved in management of those CICSplexes. As most of the CICSPlex SM definitions required for MAS resource monitoring can be reused for analysis point monitoring, the cost of staging the implementation in this way is not high. For example, a single action definition is likely to be usable by multiple analysis definitions. However, for analysis point monitoring you must create at least one analysis point specification. As in the case of MAS resource monitoring, you should start analysis point monitoring with simple definitions, and move gradually to using more complex definitions.

A recommended approach

When you decide to implement RTA functions in your enterprise, you should begin by defining the appropriate objects. You should:

Then, to create associations between these objects:

You can use the MAP action command to display a visual map of the real-time analysis definitions in your data repository. If you want to see a list of the objects and associations you have already defined, use the RTAINGRP, RTAINSPC, and RTASCOPE views.

CICSPlex SM includes a Starter Set of CICSPlex SM object definitions that illustrates the concepts and components introduced in this chapter. The Starter Set is described in detail in The CICSPlex SM Starter Set.

If you have problems defining any particular object or association, refer to the CICSPlex SM Starter Set examples. The Starter Set includes many real-time analysis examples that you can use as models for your own objects. You can copy the supplied samples directly into your data repository via the batched repository-update facility. For more information about using the Starter Set as a model and about the batched repository-update facility, see the CICS Transaction Server for z/OS® Installation Guide and CICSPlex System Manager Administration, respectively.

[[ Contents Previous Page | Next Page Index ]]