Profiles for Looping Dependent Regions

The Profiles for Looping Dependent Regions provide different types of profiles that you can use to create alarms to monitor looping IMS regions.

For each profile type, you can create one or more profiles based on different loop criteria that you define. The defined profiles can then create alarms to detect dependent region loops based on total transaction CPU, CPU usage percentage, DL/I and SQL calls rates, and I/O rates.

With the Profiles for Looping Dependent Regions, you can define alarms to monitor for:

• A never-ending loop
  By specifying a minimum transaction CPU trigger value and a CPU percentage value, you can indicate when to start looking at the region for a possible loop and the percentage of CPU consumption that you feel is indicative of a loop. In addition, specifying 0, with operand =, in the I/O, DL/I, and SQL rate fields helps to identify tight loops that do not result in these kinds of events.

• A loop involving DL/I, I/O, DL/I calls, or SQL calls
  You can detect these types of loops by specifying the >= operand in the DL/I, I/O, or SQL calls fields in the profile's loop criteria. You should also consider whether to specify a CPU usage trigger, as it might not be necessary to detect the loop. Alternatively, you could create alarms to monitor for excessively high rates for these categories. 
  
  
• Excessive total CPU used
  In the loop criteria, specify a transaction CPU trigger value that defines the excessive amount. Depending on your needs, you might or might not want to specify values in the other loop criteria fields.

Depending on the loop criteria that you define, you can use the profiles to create alarms that can monitor IMS regions for the following conditions:

• Excessive CPU time used
• High CPU percentage
• High I/O rate
• High DL/I call rate
• High SQL call rate
• Some combination of the listed metrics

You can also specify a specific or generic IMS job name for the profile and make specifications to include or exclude these items:

• IMS application region job names
• Region types
• Transaction names
• LTERM names
• Program names

Each profile type provides you with the ability to create alarms to either issue messages without taking any actions or to issue messages and execute a series of actions to address any problems that are detected. If desired, the alarms can perform the following actions:

• Stop the looping IMS region with an ABDUMP command
• Cancel the IMS region if the ABDUMP remains pending after a specified time
• Quiesce the IMS region by using the RESET command to change the WLM service class

The profiles create alarms that can work together to remedy looping IMS regions. For instance, the first profile creates an alarm that can detect a looping IMS dependent region and issue a STOP REGION ABDUMP command, which might resolve the immediate problem by stopping application with a U0474. If the loop persists after the ABDUMP, a second type of alarm can reset the WLM service class to quiesce the region–—if the alarm determines that a cancel would result in a U0113.  A third type of alarm can issue a RESET Service Class when the loop is persisting for 90 seconds. The reset action can quiesce the looping region by putting it in a lower WLM service class. It will remain quiesced until you stop IMS or otherwise cancel the region.  

This section contains the following topics: