One the questions often asked is "will zIIP save me money?" (there are other reasons to exploit zIIP - but this seems to be the biggest one). As I mentioned in my CA-WORLD presentation - the only way to answer this question is: "does the amount of TCB-mode (billable) cpu go DOWN under zIIP exploitation?
[color=#1f02b0]Here is a job stream that may help determine the answer - you might consider running something like this every night at shutdown for a week (or month) BEFORE and AFTER z/iip exploitation on V18[color]
Chris hoelscher
Database Administrator |Technical Services Humana
123 East Main Street |Louisville, KY 40202 T 502.476.2538 F
choelscher@humana.com Humana.com Keeping CAS and Metavance safe for all HUMANAty
1) RHDCUCFB
DCMT D SUBTASK
BYE
2) Parse the output from (1) to create the following
DCMT D SUBTASK 01
DCMT D SUBTASK 02
DCMT D SUBTASK 03
etc
For as many subtasks as you have - in CULPRIT - this would be written as
3) RHDCUCFB and
DCMT D TIME
DCMT D STAT SYS
DCMT D SUBT EFF
followed by the output of (2)
4) Parse the output from (3) into a save-able file - i just happen to have a culprit to do just that (surprized?)
[color=#1f02b0]( ... see attachment ... )[color]
the output file can be save with previous runs from the same CV or from previous CVs
5) then print useful? information from the accumulateD FILE (4)
[color=#1f02b0]( ... see attachment ... )[color]
Obviously many things can be reported on - here are just a few of the things I would look for from the data
[color=#2802e2]
Run this for a reasonable amount time BEFORE and AFTER turning on ziip in v18 (this report is calibrated to V18) The bottom line is ... if billable per task goes down AFTER turning ziip on - ziip is most likely a good idea If billable per task goes up (and that could happen only if the cpu/task more than doubles) - then ziip is NOT a good idea
Feel free to contact me with any questions on this mess of code I have provided[color]