My first reaction to this suggestion was "as if". Then I remembered from long ago an actual IBM recommendation that TSO 'performance expectation' be kept in check by deliberately setting response to some defined level like (say) three seconds. The idea was to keep response time steady whether the machine was lightly or heavily loaded. So users would not complain. I was infuriated by this recommendation and never dreamed of implementing it, but some shops may have. Still, I'm sticking with "as if".
. . J.O.Skip Robinson Southern California Edison Company Electric Dragon Team Paddler SHARE MVS Program Co-Manager 323-715-0595 Mobile 626-543-6132 Office ⇐=== NEW [email protected] -----Original Message----- From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf Of Rugen, Len Sent: Wednesday, January 10, 2018 7:34 PM To: [email protected] Subject: (External):Re: System z & Meltdown/Spectre I wonder if the microcode has a little "detuning" so that if they ever have to make patches that would slow down the system, they can remove a little of the detuning to get back to base performance. Len Rugen University of Missouri Division of Information Technology Systems & Operations - Metrics & Automation Team ________________________________ From: IBM Mainframe Discussion List <[email protected]> on behalf of Mike Schwab <[email protected]> Sent: Wednesday, January 10, 2018 8:55:07 PM To: [email protected] Subject: Re: System z & Meltdown/Spectre Yep. I know my site didn't purchase upgrades until running at 100% all weekday long would slow down batch runs by 4X elapsed time compared to 90%. On Wed, Jan 10, 2018 at 1:15 PM, Pommier, Rex <[email protected]> wrote: > Gil, > > I don't think this one is necessarily a quality issue. As widespread as the > problem is, it appears to me to be more of a huuuge oversight. That said, > you are correct in that we don't know what IBM plans to do, or for that > matter, even how much of an impact fixing this will be on performance of > z/OS. My concern is that since we (as well as probably many of the sites on > this board) are bumping against our cap and if we get hit by even a 10% > reduction in thruput from this, we're in danger of missing SLAs and other fun > TLAs. > > Rex > > -----Original Message----- > From: IBM Mainframe Discussion List [mailto:[email protected]] > On Behalf Of Paul Gilmartin > Sent: Wednesday, January 10, 2018 11:52 AM > To: [email protected] > Subject: Re: System z & Meltdown/Spectre > > On Wed, 10 Jan 2018 17:31:37 +0000, Pommier, Rex wrote: > >>And what does "a significant CPU hit" mean for those of us who's 4HRA is >>already at the top of our machine's capacity? That's where my concern is. >>Yeah, the money for increasing MSUs is gonna stink, but missing SLAs due to >>the box no longer being big enough is going to really bite. >> > There's a cynical suspicion that when a vendor sees defect repair as a > profit center it provides a disincentive to quality. > > This might go a step beyond: "We did it wrong, so we'll raise the price to > make it right." > > In fairness, we don't know what IBM plans to do. > > -- gil -- Mike A Schwab, Springfield IL USA ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
