Performance-minded customers are always looking to ensure they get the best value out of their hardware. Since their introduction in 2005, customers have been utilizing zIIP engines to offload processing from general purpose (GP) CPs to avoid MIPS consumption. Given the sheer number of CICS transactions performed in a shop, it's logical for customers to request zIIP offload for CICS-based transactions. However, gathering this data is not so easy.
Release : 16.0
Component : SYSVIEW
At the time of this writing, IBM doesn't provide separate entries for determining zIIP offload when processing CICS transactions. SYSVIEW derives much of its CICS transaction data from native CICS control blocks, as such there is not native ziip time available. However, IBM documentation provides the calculations needed to calculate the transaction zIIP offload time.
Viewing a CTRANLOG log, the "Degradation Analysis" section contains a set of CPU-related values. There are definitions for CPU-related fields but this aata doesn't include any zIIP-specific metrics. Doing some additional research, I found the following IBM article that discusses how to calculate this information:
"To calculate the task processor time that was spent on a specialty processor (zIIP or zAAP), subtract the time recorded in the CPUTONCP field from the time recorded in the USRCPUT field."
Using Sysview, this can be done using the CTRANLOG "DFHGroup - DFHTASK" entries:
Report Writer Sample:
TITLE1 'CICS ZIIP time'
DEFINE VAR CICS TRAN ZIIPTM = ,
CICS TRAN C_USRCPUT_TM - ,
CICS TRAN C_CPUTONCP_TM
FLASHBACK CICS DATETIME TRANID,
CICS TRAN C_USRCPUT_TM AND,
CICS TRAN C_CPUTONCP_TM AND,
CICS TRAN ZIIPTM FORMAT(NN.NNNNNN)