Hi, generally speacking about standard JOBS/STC step end (SMF30 subtype 4) I usually use:
smf30cpt + smf30cps + smf30isb + smf30icu + smf30hpt + smf30iip + smf30rct for CPU on standard CP. SMF30ISB Initiator CPU time under the service request block (SRB), in hundredths of a second. This field is set at step termination SMF30ICU Initiator CPU time under the task control block (TCB), in hundredths of a second. This field is set at step termination. SMF30HPT CPU time consumed for the step, in hundredths of a second, to support requests for data to be transferred between a hiperspace and an address space, when the hiperspace is backed by expanded storage. The CPU time may vary depending on the availability of expanded storage SMF30IIP Amount of CPU time used to process I/O interrupts, in hundredths of a second. SMF30RCT Amount of CPU time used by the region control task (RCT), in hundredths of a second. SMF30CPS Step CPU time under the service request block (SRB), in hundredths of a second. You can calculate the SRB time in microseconds (1/100 of a second using the following formula: (SMF30SRB*10)/SMF30SRC * SMF30SUS/16 = SRB time in microseconds. This value includes the value in field SMF30OST SMF30CPT All standard CPU step time in hundredths of a second. Includes enclave time, preemptable class SRB time, client SRB time. Also includes time consumed by zAAP or zIIP eligible work running on a standard processor. This value includes the value in field SMF30OST Also: smf30_time_on_ziip to have zIIP cpu usage and smf30_time_ziip_on_cp to have zIIP eligible CPU ran on standard CP because no zIIP processor was available. It's not as good as RMF report but it can be considered a goog approx. Best regards. Max Il giorno mer 5 set 2018 alle ore 11:32 Elardus Engelbrecht < elardus.engelbre...@sita.co.za> ha scritto: > To all zIIP gurus, > > I have read Martin Packer's SHARE presentation and the thread 'Why are > highly busy zIIPs worse than highly busy CPs?' in June 2018 amongst other > things like Google searches, SMF and RMF books. > > I want to know how can I see (from SMF records 30 according to Martin > Packer and others), how much CPU was used in total by an address space. > > Something like a total of SMF30CPT, SMF30CPS, SMF30HPT plus > SMF30_TIME_ON_zIIP? > > To what all fields should I look so I can get the total time (actual work > done and delays) spend on general CPU and zIIP? I am sure I have missed > useful fields which I can use? > > Should I also consider the fields about delays too? > > What is 'independent enclave time' and 'derived processor times'? > > Many thanks in advance. > > PS: I did read in the SMF book where it is described partially how CPU and > SRB times are calculated. > PS: I am not interested how WLM and RMF is reporting on usage and > scheduling of zIIP. I got that info. I just need the info PER address space. > > Groete / Greetings > Elardus Engelbrecht > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN > ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
