Re: SMS Compression measurment?
Not worried about the efficiency of compression. The issue is showing to someone the CPU saved by using zEDC instead of SMS compression. IBM's zBNA tool estimates it, and we may just use that, but I was hoping someone had a way to measure the CPU used just for SMS compression in a job, without having to run it twice. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
Re: SMS Compression measurment?
On Mon, 29 Apr 2019 11:32:34 -0500, Tim Hare wrote: >Yes, I've looked at Scott Barry's presentation. It tells me a lot of good >info about zEDC measurements, and it does document compression information in >type 14/15 but it appears as though IBM didn't provide any measurement of CPU >used by SMS compression. I guess the fallback is to try to run the same job >twice, once with compressed output and once without compression (and also read >each back in). As I see it, the consideration needed is that any file is going to "data/content dependent", therefore the idea of simulated/estimation for SMS compression would be questionable, that is without actually processing the file -- the method used with clients we support is to perform a BEFORE / AFTER with several data-samples. Also, the compression-method would also be a factor in terms of compression-result efficiency. Scott Barry SBBWorks, Inc. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
Re: SMS Compression measurment?
Yes, I've looked at Scott Barry's presentation. It tells me a lot of good info about zEDC measurements, and it does document compression information in type 14/15 but it appears as though IBM didn't provide any measurement of CPU used by SMS compression. I guess the fallback is to try to run the same job twice, once with compressed output and once without compression (and also read each back in). -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
Re: SMS Compression measurment?
Scott Barry's Share Session 24421 in Phoenix this Spring provides a good analysis of zEDC analysis, including this list of relevant SMF records. Here are the various SMF/DCOLLECT record references: - RMF 74 subtype 9 - SMF type 14/15 - SMF type 30, zEDC data-section - SMF type 23 (for SMF Logstreams and zEDC compression) - SMF type 88 (again, for SMF Logstreams / compression) - DCOLLECT – record types “D” (Dataset/VTOC), “M” (Migrated) and “B” (Backup). Barry Merrill Herbert W “Barry” Merrill, PHD President-Programmer Merrill Consultants MXG Software 10717 Cromwell Drive Dallas, TX 75229 www.mxg.com 214 351 1966 ad...@mxg.com for business questions supp...@mxg.com for technical questions ba...@mxg.com -Original Message- From: IBM Mainframe Discussion List On Behalf Of Tim Hare Sent: Monday, April 22, 2019 3:24 PM To: IBM-MAIN@LISTSERV.UA.EDU Subject: SMS Compression measurment? I can't remember if I posted this before, but it's come up again so I'm asking. Is there any way, short of running a job with SMS turning compression, and running it a different time with compression off, to measure the CPU time used by SMS compression for a dataset? I know I can get zEDC CPU savings estimates from zBNA and I'm pursuing that path too, but it seems to me it's something that ought to be reported; especially since recent versions of z/OS have zEDC measurements reported to SMF. -- 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
SMS Compression measurment?
I can't remember if I posted this before, but it's come up again so I'm asking. Is there any way, short of running a job with SMS turning compression, and running it a different time with compression off, to measure the CPU time used by SMS compression for a dataset? I know I can get zEDC CPU savings estimates from zBNA and I'm pursuing that path too, but it seems to me it's something that ought to be reported; especially since recent versions of z/OS have zEDC measurements reported to SMF. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
