Dave Heggen - 10 December 2013

IBM APAR OA39993 available for the zEC12 processor introduced what could be considered a new Service Time component for devices, Interrupt Delay Time.  This value measures the time from when the IO completes to when the z/OS issues Test SubChannel (TSCH) to view the results.

Not new, but previously Uncaptured

I selected the title for this Blog as “nil sub sole novum”. It’s a common Latin phrase whose literal translation is ‘Nothing under the Sun is new’, meaning ‘everything has been done before’.  This is very true for Interrupt Delay Time, the component is new but the activity has always been with us. This activity was previously without description.  It came from what you could call ‘Uncaptured IO Time’.  Even my spell checker rebels against the use of the word uncaptured.

The Anatomy of an IO

Anatomy of an IO

Uncaptured time in the processor is a familiar statistic.  Uncaptured IO Time is the discrepancy between what RMF would report response time to be and what an application (like CICS) could measure for response time for the same IO.   In the above chart, the two arrowed labels containing the word ‘Path’ are still uncaptured IO time (slots 1 and 6).

Uncaptured IO time now Captured

So Interrupt Delay Time (or IO Delay Time) is a measure of IO Elongation (the fifth slot from the left in the picture above).  If high values are presented in this field, the possible causes might be a non-Dispatched LPAR or the lack of an interrupt enabled CP.

Conclusion

Support for tracking Interrupt Delay Time is provided in IntelliMagic Vision Version 7.  By default, Warning and Exception Flags are raised in IntelliMagic Vision when Interrupt Delay times are >.3 and >.6 milliseconds, respectively.  The expected value of this component is less than 0.1 milliseconds.  Should this value become troublesome because of LPAR Dispatch Delay, a more aggressive value can be used for CPENABLE parameter in the IEAOPTxx parmlib member by changing it from the default value of (10,30) to something like (15,20).

Breaking out the Interrupt Delay Time from the uncaptured IO time is first possible on the IBM zEC12/zBC12 processors and applies to any vendor’s storage equipment.  I think we can all agree that Interrupt Delay Time is a timely addition as we embrace sub-millisecond response time reporting.

Related

Webinar

Understanding & Dealing with z14 Traffic Patterns

The z14 is designed for massive, parallel processing. So why do delays still occur? This webinar will explore common sources of application delays and discuss practical solutions to reduce these delays.

Watch Webinar
Webinar

Modernize Db2 SMF Analysis with Smarter Algorithms

The Db2 SMF data is a rich and voluminous source of metrics, but the size and complexity make it difficult to know when the metrics start to get into conditions that represent problems, risk, and best-practice violations.

Watch Webinar
Blog

IntelliMagic Sessions at IDUG Charlotte 2019

After releasing our support for Db2 for z/OS earlier this year, we decided to sponsor the North American IDUG conference and the response was overwhelmingly positive.

Read more

Go to Resources

This article's author

Dave Heggen
Storage Performance Consultant
Read Dave's bio

Share this blog

Subscribe to our Newsletter

Subscribe to our newsletter and receive monthly updates about the latest industry news and high quality content, like webinars, blogs, white papers, and more.