Typical performance review questions

1. Has the frequency of use of each transaction identifier altered?
2. Does the mix vary from one time of the day to another?
3. Should statistics be requested more frequently during the day to verify this?

In these cases, you must identify the function by program or data set usage, with appropriate reference to the CICS program statistics, file statistics, or other statistics. In addition, you might be able to put user tags into the monitoring data (for example, a user character field in the case of the CICS monitoring facility), which can be used as a basis for analysis by products such as CICS Performance Analyzer for z/OS, or IBM Z Decision Support.

1. Do the CICS terminal statistics indicate any increase in the number of messages on the terminals on each of the lines?
2. Does the average message length on the CICS performance class monitor reports vary for any transaction type? This can easily happen with an application where the number of lines or fields output depends on the input data.
3. Is the number of terminal errors acceptable? If you are using a terminal error program or node error program, are there any line problems?

1. Is the number of requests to file control increasing? Remember that CICS records the number of logical requests made. The number of physical I/O operations depends on the configuration of indexes, and on the data records per control interval and the buffer allocations.
2. Is intrapartition transient data usage increasing? Transient data involves a number of I/O operations depending on the queue mix. Review the number of requests made to see how it compares with previous runs.
3. Is auxiliary temporary storage usage increasing? Temporary storage uses control interval access, but writes the control interval out only at sync point or when the buffer is full.

1. How large are the dynamic storage areas?
2. Is the number of GETMAIN requests consistent with the number and types of tasks?
3. Is the short-on-storage (SOS) condition being reached often?
4. Have any incidents been reported of tasks being purged after deadlock timeout interval (DTIMOUT) expiry?
5. How much program loading activity is there?
6. From the monitor report data, is the use of dynamic storage by task type as expected?
7. Is storage usage similar at each execution of CICS?
8. Are there any incident reports showing that the first invocation of a function takes a lot longer than subsequent ones? This situation can occur if programs are loaded that then need to open data sets, particularly in IMS, for example. Can a change in application design rectify the problem?

1. Is the processor usage as measured by the monitor report consistent with previous observations?
2. Are batch jobs that are planned to run, able to run successfully?
3. Is there any increase in usage of functions running at a higher priority than CICS? Include MVS readers and writers, MVS JES, and z/OS Communications Server if running above CICS, and overall I/O, because of the lower-priority regions.

1. What is the average storage usage?
2. What is the link utilization?

1. Are any of the resources heavily used? If so, was this situation expected at design time? If not, can the heavy usage be explained in terms of heavier usage of transactions?
2. Is the heavy usage associated with a particular application? If so, is there evidence of planned growth or peak periods?
3. Are browse transactions issuing more than the expected number of requests? In other words, is the count of browse requests issued by a transaction greater than what you expected users to cause?
4. Is the CICS CSAC transaction (provided by the DFHACP abnormal condition program) being used frequently? If so, is this occurring because invalid transaction identifiers are being entered? For example, errors are signaled if transaction identifiers are entered in lowercase on IBM® 3270 terminals but automatic translation of input to uppercase has not been specified.

   A high use of the DFHACP program without a corresponding count of CSAC could indicate that transactions are being entered without correct operator signon. This situation might indicate that some terminal operators need more training in using the system.