Planning your monitoring schedule

A comprehensive monitoring plan includes the scheduling of various system activities at different time intervals. This approach provides a broad collection of data to measure and analyze the performance your CICS® system. Plan for both dynamic monitoring and scheduled monitoring.

Dynamic monitoring

Dynamic monitoring is “on-the-spot” monitoring that you can carry out at all times. This type of monitoring includes the following activities:
  • Observing the operation of the system continuously to discover any serious short-term deviation from performance objectives. End-user feedback is essential for this activity. You can also use the Resource Measurement Facility™ (RMF™) to collect information about processor, channel, coupling facility, and I/O device usage.
  • Obtaining status information. You can get status information about system processing during online execution. This information might include the queue levels, active regions, active terminals, and the number and type of conversational transactions. You can get this information with the aid of an automated program started by the master terminal operator. At prearranged times in the production cycle (such as before scheduling a message, at shutdown of part of the network, or at peak loading), the program can capture the transaction processing status and measurements of system resource levels.
  • Using CICSPlex® SM monitoring data. CICSPlex SM can accumulate information produced by the CICS monitoring facility to assist in dynamic monitoring activities. The data can then be immediately viewed online, giving instant feedback on the performance of the transactions. CICS monitoring must be active for CICSPlex SM to collect CICS monitoring information.

Daily monitoring

Measure and record key system parameters by monitoring data daily. The daily monitoring of data usually consists of counts of events and gross level timings. In some cases, the timings are averaged for the entire CICS system. To monitor data daily, perform a series of tasks. For example:

  • Record both the daily average and the peak period (usually one hour) average of items such as messages, tasks, processor usage, I/O events, and storage used. Compare these events against your major performance objectives and look for adverse trends.
  • List the CICS-provided statistics at the end of every CICS run. Date-stamp and time-stamp the data that is provided, and file it for later review. For example, in an installation that has settled down, you might review daily data at the end of the week; generally, you can carry out reviews less frequently than collection, for any one type of monitoring data. If you know there is a problem, you might increase the frequency; for example, reviewing daily data as soon as it becomes available.
  • Be familiar with all the facilities in CICS for providing statistics at times other than at shutdown. The main facilities are invocation from a terminal (with or without reset of the counters) and automatic time-initiated requests.
  • File an informal note of any incidents reported during the run, including, for example, a shutdown of CICS that causes a gap in the statistics, a complaint from your users of poor response times, a terminal going out of service, or any other significant item. These notes are useful when reconciling disparities in detailed performance figures that might be discovered later.
  • Print the system console log for the period when CICS was active, and file a copy of the console log in case it becomes necessary to review the CICS system performance in the light of the concurrent batch activity.
  • Run one of the performance analysis tools described in Performance measurement tools for at least part of the day if there is any variation in load. File the summaries of the reports produced by the tools you use.
  • Transcribe onto a graph any items identified as being consistently heavily used in the post-development review phase.
  • Collect CICS statistics, monitoring data, and RMF data into the Tivoli® Decision Support database.

Weekly monitoring

Periodically collect detailed statistics on the operation of your system for comparison with your system-oriented objectives and workload profiles. To monitor data weekly, perform these steps:
  • Run the CICS monitoring facility with performance class active, and process it. You might not need to run the facility every day, but it is important to do it regularly and to keep the sorted summary output and the detailed reports. Whether you run the facility on the same day of the week depends on the nature of the system load. For example, if one day of the week has a heavier system load than others, monitor on this day. Bear in mind, however, that the use of the monitoring facility causes additional load, particularly with performance class active.
  • If the load is apparently the same each day, run the CICS monitoring facility daily for a period sufficient to confirm the load. If there really is little difference from day to day in the CICS load, check the concurrent batch loads in the same way from the logs. Checking the batch loads helps you identify any obscure problems because of peak volumes or unusual transaction mixes on specific days of the week. The first few weeks of output from the CICS statistics also provide useful information.You might not need to review the detailed monitor report output every time, but always keep this output in case the summary data is insufficient to answer questions raised by the statistics or by user comments. Label the CICS monitoring facility output and keep it for an agreed period in case further investigations are required.
  • Run RMF, because this shows I/O use, channel use, and other uses. File the summary reports and archive the output information for some agreed period.
  • Review the CICS statistics, and any incident reports.
  • Review the graph of critical parameters. If any of the items is approaching a critical level, check the performance analysis and RMF output for more detail.
  • Tabulate or produce a graph of values as a summary for future reference.
  • Produce weekly Tivoli Decision Support or CICS Performance Analyzer reports.

Monthly monitoring

Monitor and assess trends that are better reflected when tracked regularly over a longer period of time. The following list includes some tasks for monitoring data on a monthly basis:

  • Run RMF.
  • Review the RMF and performance analysis listings. If there is any indication of excessive resource usage, follow any previously agreed procedures (for example, notify your management), and do further monitoring.
  • Date-stamp and time-stamp the RMF output and keep it for use in case performance problems start to arise. You can also use the output in making estimates, when detailed knowledge of component usage might be important. The RMF output provides detailed data on the usage of resources within the system, including processor usage, use of DASD, and paging rates.
  • Produce monthly Tivoli Decision Support reports showing long-term trends.

Monitoring for the future

When performance is acceptable, establish procedures to monitor system performance measurements and anticipate performance constraints before they become response-time problems. Exception-reporting procedures are a key to an effective monitoring approach. In a complex production system there is often too much performance data for it to be comprehensively reviewed every day. Key components of performance degradation can be identified with experience, and those components are the ones to monitor most closely. Identify trends of usage and other factors (such as batch schedules) to aid in this process.



dfht32e.html | Timestamp icon Last updated: Monday, 5 June 2017