In real time, how determine what IMS trans are consuming most CPU

book

Article ID: 129857

calendar_today

Updated On:

Products

CA SYSVIEW Performance Management NXBRIDGE - SYSYVIEW/ENDEVOR

Issue/Introduction



A lot of times during the day, we would be alerted that the CPU utilization is high and experience latency system wide(IMS MPPs, BMPS, batch jobs, TSO etc)
he alert is sent when the SYSVIEW PARTINFO TOTAL BUSY% is greater than 90%;
Sample screen below: PARTINFO W02P01 ? -----------------
Name Busy% TOTAL 98.6 OVERHEAD 0.4 SUBTOTAL 89.2
MVS1 39.5
MVS2 4.8
MVS3 5.1
MVS4 1.7
MVS5 38.0
SUBTOTAL 9.1
CFY 9.1
Most of the time, according to SYSVIEW ACTIVITY screen, it' the IMS MPP regions that are the heaviest CPU users.
At that point, we know it's the MPPs that are using the CPU but we do not specifically know what transactions.
Questions we get from our client are:
1. What transaction(s) are consuming the most CPU from X time to Y time? IMSTIMES can provide the average lifetime, response time, MPPCPU etc but it does answer that question above.
2. What are the average lifetimes of transaction XYZ from X to Y time every Z secs(or mins or even hour)?
IMSTLOG can provide individual times from X to Y time but it will not give the average lifetime time every Z seconds(or mins or hour).
Can SYSVIEW provide these information in (near) real time?
We have SYSVIEW 15.0

Environment

Release:
Component: SYSVW

Resolution

It is not possible to create an IMSID group across LPARs.
However, it is possible to create a REXX exec that issues the IMSTIMES XSYS command, merges the transactions, and displays the output.