Re: Region size for batch processing
Thanks for all of your responses,, we going to use a region size of 1024M for most of our batch,, the exception being the SAS which or sysprogs want us to leave running with 0M. I think we still may be at risk, but it's not my call. On 12/04/06, andy corpes [EMAIL PROTECTED] wrote: Hi All, We are having discisisons regarding an optimal region size for our standard batch processing. Currently, we use region=0M for all of our batch jobs without any noticeable problems. Is this subparm useful anymore? it would appear that our system does not care too much, but i am concerned about an slowly increasing number of JAVA programs that have required up to 128meg to execute What is the groups opinion on this issue. -- Andy -- Andy -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Region size for batch processing
our sysprogs discourage (prevent?) 0M on the belief (fact?) that if such a job were to run-away and consume all available memoryon its way to ebending, there would be no room to set up abend-related control blocks in memory when it did abend - thus we are constrained to use 2047M Chris Hoelscher IDMS DB2 Database Administrator Humana Inc 502-710-3038 [EMAIL PROTECTED] The information transmitted is intended only for the person or entity to which it is addressed and may contain CONFIDENTIAL material. If you receive this material/information in error, please contact the sender and delete or destroy the material/information. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Region size for batch processing
When I was in support many years ago, many times customers would call asking us about an S40D or S0DC abends - it just went away! (For the peanut gallery, the problem is that there is no more left in below-the-line private to acquire LSQA for the RTM control blocks.) I've lost track of the times I've told people why 0M is a bad idea for a server-type program, and in general is a bad idea. A couple of weeks ago, I had to explain this to one of our developers, who has gray hair/beard (but I don't know about the hairy chest). Later, Ray -Original Message- From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of Chris Hoelscher Sent: Wednesday April 12 2006 05:47 our sysprogs discourage (prevent?) 0M on the belief (fact?) that if such a job were to run-away and consume all available memoryon its way to ebending, there would be no room to set up abend-related control blocks in memory when it did abend - thus we are constrained to use 2047M -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Region size for batch processing
've lost track of the times I've told people why 0M is a bad idea for a server-type program, and in general is a bad idea. IBM has shipped 0M for 'server-types', for years. VTAM, NETVIEW, IMS, DB2, MQ-Series, to name a few. I don't have a problem with that. Where I have a problem is with application programmes using it. At one place we had an exit (UJI?), that over-rode regular Batch and TSO to a non-zero region specification. If you needed more, you could specify it. You just couldn't specify 0M. Also, if you specify anything over 16M, you still have the same problem below the line that you have with 0M. The cushion (IEALIMIT) is gone. - -teD O-KAY! BLUE! JAYS! Let's PLAY! BALL! -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Region size for batch processing
I have a SAF call in IEFUSI, you get over 256M if you DESERVE it. I think you can address the under 16M vs. 0M limit in IEFUSI as well, I think you could set the region size to a suitable value less than GDAPVTSZ. -Original Message- snip I don't have a problem with that. Where I have a problem is with application programmes using it. At one place we had an exit (UJI?), that over-rode regular Batch and TSO to a non-zero region specification. If you needed more, you could specify it. You just couldn't specify 0M. Also, if you specify anything over 16M, you still have the same problem below the line that you have with 0M. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Region size for batch processing
andy corpes wrote: Hi All, We are having discisisons regarding an optimal region size for our standard batch processing. Currently, we use region=0M for all of our batch jobs without any noticeable problems. IMHO it's like lack of fuse in a circuit. Fuse is not needed ...unless something goes wrong. Is this subparm useful anymore? it would appear that our system does not care too much, but i am concerned about an slowly increasing number of JAVA programs that have required up to 128meg to execute What is the groups opinion on this issue. My humble opinion is you should specify any finite value. It can be big, but finite. You can even specify it in JESPARM as default, just to forget about the problem. It is no waste of resources, since REGION does not imply memory allocation, just potential limit. Very different from SPACE parameter. Most legacy programs have static memory allocation, so they won't use higher limit. Some, i.e. DFSORT can use higher limit giving better performance. However no limit does not sound reasonable for application jobs. My $0.02 -- Radoslaw Skorupka Lodz, Poland -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Region size for batch processing
If your system is not set up to deal with very large address spaces, a malfunctioning task can cause you some serious problems. On the other hand, if you are not set up for large address spaces, you could be underutilizing you main storage and hurting overall throughput. Some argue that that is a serious problem in its own right. It is not like you can just arbitrarily limit the address space size any more. I suspect these brave new application worlds are going to *require* address space sizes far beyond what many gray haired folks would consider obscene. Even 'standard' batch often contains many sorts, and there is no better way to improve sort performance than to give them all the storage you can. Big buffers can also work minor miracles. Ideally, I vote for REGION=0m for production but not test. The reality is that testers rarely have enough information to choose an 'optimal' region size. Therefore, I guess the best compromise is to select a region size that is the largest possible on your system that does not pose undue risk to the system. That is a bit easier to sell to those that don't understand. HTH. -Original Message- From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of andy corpes Sent: Wednesday, April 12, 2006 7:29 AM To: IBM-MAIN@BAMA.UA.EDU Subject: Region size for batch processing Hi All, We are having discisisons regarding an optimal region size for our standard batch processing. Currently, we use region=0M for all of our batch jobs without any noticeable problems. Is this subparm useful anymore? it would appear that our system does not care too much, but i am concerned about an slowly increasing number of JAVA programs that have required up to 128meg to execute What is the groups opinion on this issue. -- Andy -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Region size for batch processing
ABEND40D, ABEND0F9, ... are what you can get running REGION=0M without any intervention by IEFUSI so your systems programmers are right. This has been discussed at length before search terms REGION and IEFUSI will get you tons of hits in the archives. We use an IEFUSI I updated from one Mike Loos wrote. http://www.cbttape.org/ftp/cbt/CBT425.zip Mine has less code than Mike's did so if you want to secure REGION=0M then his might be a better starting place. I just have two modes of operation REGION=0M gets you maximum that you can get safely with a buffer reserved above and below otherwise set maximum less buffer below and a 1G default above. It will be here file 518 later today probably http://www.cbttape.org/updates.htm Mike Loos gave an excellent presentation on IEFUSI at SHARE a few years back. Search the archives using REGION=0M or IEFUSI as a keyword and you will locate numerous discussions of virtual storage management. Visit the SHARE web site http://www.share.org/ for Mike's foils From SHARE in Boston - July, 2000. http://shareweb.share.org/proceedings/sh95/share00s.htm 2891 - Use of the IEFUSI Exit to Limit (or not) Region Size http://shareweb.share.org/proceedings/sh95/data/S2891A.PDF http://shareweb.share.org/proceedings/sh95/data/S2891B.PDF http://www.cbttape.org/ftp/cbt/CBT425.zip also of interest is Chris Craddock's foils from this Virtual Storage session last seen in NYC I believe 2829 - z/OS Virtual Storage Mystery Tour http://ew.share.org/client_files/callpapers/attach/SHARE_in_New_York/S28 29.pdf Best Regards, Sam Knutson, GEICO Performance and Availability Management mailto:[EMAIL PROTECTED] (office) 301.986.3574 Think big, act bold, start simple, grow fast... -Original Message- From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of Chris Hoelscher Sent: Wednesday, April 12, 2006 8:47 AM To: IBM-MAIN@BAMA.UA.EDU Subject: Re: Region size for batch processing our sysprogs discourage (prevent?) 0M on the belief (fact?) that if such a job were to run-away and consume all available memoryon its way to ebending, there would be no room to set up abend-related control blocks in memory when it did abend - thus we are constrained to use 2047M Chris Hoelscher IDMS DB2 Database Administrator Humana Inc 502-710-3038 [EMAIL PROTECTED] This email/fax message is for the sole use of the intended recipient(s) and may contain confidential and privileged information. Any unauthorized review, use, disclosure or distribution of this email/fax is prohibited. If you are not the intended recipient, please destroy all paper and electronic copies of the original message. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Region size for batch processing
In [EMAIL PROTECTED], on 04/12/2006 at 02:29 PM, andy corpes [EMAIL PROTECTED] said: Currently, we use region=0M for all of our batch jobs without any noticeable problems. It's not my dog. What is the groups opinion on this issue. I don't know about the group, but IMHO you're living dangerously. -- Shmuel (Seymour J.) Metz, SysProg and JOAT ISO position; see http://patriot.net/~shmuel/resume/brief.html We don't care. We don't have to care, we're Congress. (S877: The Shut up and Eat Your spam act of 2003) -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html