Hi Denis, I was thinking about this a bit on the way home the other day. You are obviously frustrated and none of what I suggest will be a silver bullet but there are some things you can do if you have not already.
TUNE! Tuning the most heavily CPU intensive portion of those IMS transactions might well pay for itself in a month or two. There are tools like Intune http://www.bmc.com Strobe http://www.compuware.com APA http://www.ibm.com http://www-306.ibm.com/software/awdtools/apa/ Freezeframe http://www.macro4.com and many other profile tools including inspect in OMEGAMON http://www.tivolii.com and CPU PROFILE in TMONMVS http://www.asg.com all of which are available on trials if you don't have any of them installed. GTF trace requires considerable more time to grok but it's included with z/OS for free. Once you find the most expensive program or section of a program optimize it and bring in help if you need it. This might be a case where a product like Data Kinetics tableBASE http://www.DKL.com or a custom developed high performance assembler search routine by http://www.celestini.com/ Art Celestini or one of the other available Wizards For Hire (WFH's) might be the ticket. If nothing else gnaw on the LE performance tips floating in IBM documentation, SHARE papers, and dig into the algorithm that is being used to process the data. Pull out the Knuth's if you must http://www-cs-faculty.stanford.edu/~knuth/taocp.html I have found that simply fetching the appropriate volume from the shelf and placing on the desk beside the terminal often exorcises demons from my programs:-) NEGOTIATE! MANAGE! BEAT! vendors into better pricing! Start a SAM Software Asset Management Effort and get off MIPS based pricing with vendors that matter. Participate in the ISVCOSTS mailing list for starters. You might want to join the ISVCOSTS mailing list and review the archives and pose your question in that forum. ISVCOSTS is a no-vendors-allowed discussion list for open discussion by IBM customers of ISV cost issues. http://www.ibm.com/software/solutions/isvcosts/ NOTE: the updated URL courtesy of IBM's genius web master who keeps moving popular pages without providing forwarding so http://www.can.ibm.com/isvcosts/ doesn't work anymore:-( Don't be afraid again to get some help with tooling or advice from folks who know how to play the game http://www.sherkow.com/ and maybe can help you leverage IBM's various pricing schemes like WLC charging to save some money. If all else fails closet yourself and read "What Would Machiavelli Do?" http://www.amazon.com/gp/product/0066620104 and get in touch with your inner evil prince before negotiating that next contract. A good SAM plan is probably the best way to improve the TCO of your mainframe for the business. Good Luck! Best Regards, Sam Knutson, GEICO Performance and Availability Management mailto:[EMAIL PROTECTED] (office) 301.986.3574 Now he has departed from this strange world a little ahead of me. That means nothing. People like us, who believe in physics, know that the distinction between past, present, and future is only a stubbornly persistent illusion. Albert Einstein -----Original Message----- From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of Denis Gaebler Sent: Wednesday, April 05, 2006 8:37 AM To: [email protected] Subject: Re: Looking for SRB sample in PL/I or COBOL Hi List, thanks for the responses so far. It was just an idea, if there was a possibility to move some CPU intensiv pieces of code out to the upcoming zIIP processors (requires SRB envlave mode). In our shop the main cost driver is CPU. Nevertheless, I assume there are more issues waiting, if calling routines as SRB. In high level languages it would be hardly possible to ensure the SRB restrictions such as not calling SVCs etc. Not to mention the authorization issues for normal programs. If that were possible I would have continued to investigate if SRBs could be used in IMS transactions. Some of our IMS transactions do a lot of scanning of in memory tables, which requires a lot of CPU. Since this pieces of code just do memory work, I thought that were candidates for SRBs, since those code sections do not use any IMS services or SVCs. Thanks for your responses. Denis Gaebler. ==================== 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
