LISTSERV at the University of Georgia
Menubar Imagemap
Home Browse Manage Request Manuals Register
Previous messageNext messagePrevious in topicNext in topicPrevious by same authorNext by same authorPrevious page (May 1997, week 3)Back to main SAS-L pageJoin or leave SAS-L (or change settings)ReplyPost a new messageSearchProportional fontNon-proportional font
Date:   Sun, 18 May 1997 02:51:25 GMT
Reply-To:   Keith Brown <kbrown@ARLINGTON.NET>
Sender:   "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:   Keith Brown <kbrown@ARLINGTON.NET>
Organization:   StarText department of the Fort Worth Star Telegram
Subject:   Re: MVS - Long running jobs using views against indexed SAS datasets

I don't have the books in front of me so I can't give you an exact reference, but I recall reading that the timer functions do terrible things to performance using views. Something like generating a stimer call per observation rather than a single start and stop timer macro. Try setting NOSTIMER and NOSTATS options and see if that fixes it.

On the other hand, if you are seeing more EXCPs, you might try the HSWORK option to substitute hiperspace memory accesses for the disk IO. That can have the side effect of reducing the CPU as well, but almost always cuts the elapsed time down. But if it causes paging problems, then your data is too big for hiperspace.

Good luck!

On Thu, 15 May 1997 07:04:32 -0700, Thomas Hanan <thomas.hanan@PRUDENTIAL.COM> wrote:

>I was wondering if any of the MVS SAS gurus out there had any insight into >this. Is there some reason that using a view against an indexed SAS >dataset (assuming that the view uses an indexed field) would cause jobs to >run MUCH longer (sometimes 10-20 hours longer) than running the same job >directly against the indexed files. For example, if my SAS dataset is >indexed on field X, it seems that our jobs using code like the following >run forever: > >DATA VIEW.TEMP/VIEW=VIEW.TEMP; > SET INDEXED.DATASET(WHERE=(X='SOMETHING')); >DATA TEMP2; > SET VIEW.TEMP; > WHERE Y='SOMETHING ELSE'; > >But if we simply use this code, the jobs run in a reasonable amount of >time: >DATA TEMP; > SET INDEXED.DATASET(WHERE=(X='SOMETHING' AND Y='SOMETHING ELSE')); > >The files coming in are huge, but not anything that should make these jobs >run for days. There is a reason we are using the views that I won't go >into here, but the performance (clock time, not cpu time) is leaving much >to be desired. Is there anything that can be done about this? Has anybody >else noticed similarly long-running jobs when using views? > >Thanks in advance, >Tom > >The views expressed here are my own and do not represent those of my >employer.


Back to: Top of message | Previous page | Main SAS-L page