Jesse, Each sort does optimization, not just each sort product, but each sort executed. There are a lot of factors that go into optimization like memory layout, other work running on the machine, what disk drives are available. Lots of other things also.
Now, that effects how long each intermediate string is, which in turn effects how the final merge will run. Both DFSORT and Syncsort have an EQUALS parameter, both as a job option and as a default installation option. If reproducibility of record sequences is important, then use the EQUALS option. It adds to the key length, but nothing comes for free, although it would have to be a pretty huge sort to notice the difference. I would say that most installations have EQUALS as their default. Chris Blaicher Technical Architect Syncsort, Inc. -----Original Message----- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf Of Jesse 1 Robinson Sent: Thursday, October 25, 2018 2:37 PM To: IBM-MAIN@LISTSERV.UA.EDU Subject: Re: SORT not behaving consistently I for one find this discrepancy highly disturbing unless it can be tracked down to an unintended difference in SORT options. Like it or not, a program should give consistent results on every execution wherever it's run. Whether it matters to the user in this particular case is not germane. It should matter to all of us. . . 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 robin...@sce.com -----Original Message----- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf Of Beesley, Paul Sent: Thursday, October 25, 2018 11:29 AM To: IBM-MAIN@LISTSERV.UA.EDU Subject: (External):Re: SORT not behaving consistently My thought exactly. Apparently it does... which, as Charles Mills correctly says, the rest of the fields should be part of the sort key or they should use EQUALS. Regards and thanks Paul -----Original Message----- From: IBM Mainframe Discussion List <IBM-MAIN@LISTSERV.UA.EDU> On Behalf Of Cameron Conacher Sent: Thursday, October 25, 2018 5:37 PM To: IBM-MAIN@LISTSERV.UA.EDU Subject: Re: SORT not behaving consistently So are you saying that the order of records with identical keys is different machine to machine? Does that matter? Sent from my iPhone > On Oct 25, 2018, at 12:23 PM, Beesley, Paul <paul.bees...@atos.net> wrote: > > Hi > > Have been asked to investigate reasons for an identical DFSORT behaving > differently on 2 machines. > We are transitioning a service which is currently on a zEC12 to a z14 > machine, and the disks are currently HP9500, moving to IBM DS8886. > When they run sort of multi-thousand records on the old machine, and then > running the identical sort against identical data on the new setup, they get > several records in a different order. > Seems they are sorting on the first 21 characters or so, and it is doing that > fine, but the order of the records beyond that is a bit random, whereas on > the new system they are in the same order as the input. > On the new system it's almost like EQUALS is coded, but it's not. > Hope that makes sense. > Any ideas what would cause this (repeatable) difference in the outcome ... > faster machine? Faster disks? > Paul ---------------------------------------------------------------------- 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
