Identifying peak waiting time periods

Using the Waiting Time Statistics report helps you identify the peak periods of the concurrent managers. You might encounter concurrent managers with excessive waiting times in one of the following situations:

• The peak waiting time periods of a concurrent manager occurs during specific, predictable time periods, such as the work shift. In this case you can set up additional concurrent managers or target processes and assign them to a specific work shift.
• The peak waiting time periods of a concurrent manager occurs due to specific concurrent requests assigned to it. In this case, you can assign certain requests to specific concurrent managers so they will not interfere with other requests.

The Waiting Time Statistics report displays the average waiting time for all requests that have run on all concurrent managers during the specified time interval. The report includes all defined concurrent managers.

In addition to identifying the waiting time statistics of all concurrent managers, you can review the Waiting Time Statistics report of a single concurrent manager during a specified time interval. The report includes only the requests that have run on the selected concurrent manager.

Besides the report, there is also a parameter with which you can review the waiting time statistics of a single concurrent manager. The WaitingTimeStats graph, located within the <concurrent_manager_name> (OA_MGR) application, shows average waiting times for a selected concurrent manager. This parameter displays average waiting time for requests that started running after the last performance statistics polling.

Before you identify the peak waiting periods, survey the user community to help identify long running concurrent requests or time periods where critical concurrent requests must run.

This topic contains the following information:

• • To identify peak waiting time periods for all concurrent managers
  • To identify peak waiting periods for specific concurrent managers
  • Related topics

To identify peak waiting time periods for all concurrent managers

Survey the user community to help identify long running concurrent requests or time periods where critical concurrent requests must run.

1. Follow the pathway applicationID (OA_SYSTEM) > CONC_MGR (OA_CONM) pathway.
2. From the CONC_MGR (OA_CONM) menu, select the Performance Analysis > Waiting Time Statistics command.
   The Enter Start Date and End Date dialog box appears.
3. Type the beginning and end dates for the report interval. You can delete and override the default date interval. The date must be in the DD-MON-YYYY format, with MON in uppercase format (JAN for January, FEB for February, etc.).
4. Click Accept.
   The interval for the waiting times statistics is now set.

The Waiting Time Statistics report displays the average waiting time for all requests that have run on concurrent managers during the specified time interval. The report includes all defined concurrent managers.

To identify peak waiting periods for specific concurrent managers

1. Follow the applicationID (OA_SYSTEM) > CONC_MGR (OA_CONM) > CONMGR_nodeName (OA_CONM_HOST) > concurrent_manager_name (OA_MGR) pathway.
2. From the concurrent_manager_name (OA_MGR) menu, select the Waiting Time Statistics command.
   The Enter Start Date and End Date dialog box appears.
3. Type the beginning and end dates for the report interval. You can delete and override the default date interval.
   The date must be in the DD-MON-YYYY format, with MON in uppercase format (JAN for January, FEB for February, etc.).
4. Click Accept.
   The interval for the waiting times statistics for the selected concurrent manager is now set. You can view the results by double-clicking the Waiting Time Statistics report.
   In addition, you can view the graph of the WaitingTimeStats parameter, showing average waiting times for the selected concurrent manager. This parameter is located within the concurrent_manager_name (OA_MGR) application.
   You can see a large sampling of data because you can specify the time period. The amount of data the reports return will depend on how frequently the log files are purged.

Related topics

Resolving-bottlenecks

Determining-concurrent-manager-utilization
Identifying-the-number-of-active-target-processes
Identifying-running-and-pending-requests
Identifying-long-running-and-long-pending-requests
Determining-concurrent-request-history
Alleviating long-running and long-pending requests