Now after 2 years since I posted this item / concern, I see numerous VIEWs but no one has commented whether or not they also have concern that CSUAVTTM (CICS average response time) is flawed for CICS/MRO 'application transaction' reporting -- unless action / awareness is taken to capture the TOR (now identified as originating transaction code, MICS CICCSU detail variable CSUOTRAN) and only count CSUTRSTM observation values for this TOR-initiated transaction.
An oversimplified example is shared in the Excel screen-capture below:
The multi-region CICS application transaction execution of "APPX" will report a response-time of 0.83 seconds, when the MICS CICCSU SAS variables CSUTRSTM and CSUTRANS are summarized in order to derive CSUAVTTM (metric in blue at the bottom: sum-of CSUTRSTM divided by sum-of CSUTRANS, for the three related transactions), otherwise a more accurate application transaction response time (as experienced by the end-user, or otherwise 'internal response') would be the TOR-only CSUTRSTM (or CSUAVTTM, same value).
One possible solution is to capture CSUOTRAN as well as TRANCODE with a CICACTn (CICS Account Code) variable and interrogate the CICCSU at reporting, resetting CSUTRSTM where CSUOTRAN does not equal TRANCODE.
At this point in time, CA has agreed to investigate further....
Scott Barry
SBBWorks, Inc.