Well, now you've top posted so I have to do the same thing. :-) I think you were confirming with what I wrote to John and I guess I can see that being true (5% # of lines of code vs. resources). But that small percent (5% or not) is the part you have to pay most attention to. You don't have much if any control of the other parts.
So saying it should be ignored because it's a small percentage of the number of lines of total code running around the system is absolutely the wrong thing to do. My client does constant analyzing of application code (batch, CICS, WebSphere) and has made a huge difference to CPU consumption numbers in many cases (some small gains, some big ones). This usually happens in testing, but sometimes is missed. There are major application releases every other month and minor ones every month and despite all the testing, we still run into unexpected utilization increases at times. Without analyzing and changing code to address what has been found in testing and sometimes after it hits production, there would have already been many more millions of dollars spent on CPU upgrades. Small amount of user code or not, you should "worry about the optimisation of the small stuff". And yes, it is the fault of the programmer or bad indexes or whatever (but not the OS's fault) - so what? If you have this same discussion for a server running wintel or unix, the same is probably more true. It's all the OS and infrastructure in terms of number of lines of code, but it is still the business application that will make the box over utilized or more utilized than a well written app. Mark -- Mark Zelden - Zelden Consulting Services - z/OS, OS/390 and MVS mailto:[email protected] Mark's MVS Utilities: http://home.flash.net/~mzelden/mvsutil.html Systems Programming expert at http://expertanswercenter.techtarget.com/ On Fri, 25 Jun 2010 20:53:51 +0000, Ted MacNEIL <[email protected]> wrote: >I don't even know how to subset this message to formulate a response. >So, I'm replying intact. > >But, my point is, while the percentage may not believed, how many users write access methods, terminal handlers, abend handlers, DB2 optimisers, ENQ-Handlers, operator commands, interupt handlers, message passers, or any other system function. > >These functions are requested by programmes, operators, terminal users, or whatever. > >Empirical data has shown, to me at least, including statements from IBM, that the user code is a small part of this. > >So, what is the point, on today's fast machines, about worrying about the optimisation of the small stuff? > >Yes, you can drive a z/Box nuts with poorly indexed DB's or bad loops. >But, are these faults caused by the language, or bad programmers, DBA's, or others? > >- >Too busy driving to stop for gas! > >-----Original Message----- >From: Mark Zelden <[email protected]> >Sender: IBM Mainframe Discussion List <[email protected]> >Date: Fri, 25 Jun 2010 15:33:05 >To: <[email protected]> >Reply-To: IBM Mainframe Discussion List <[email protected]> >Subject: Re: Delete all members of a PDS that is allocated > >On Fri, 25 Jun 2010 15:25:58 -0500, McKown, John ><[email protected]> wrote: > >>> -----Original Message----- >>> From: IBM Mainframe Discussion List >>> [mailto:[email protected]] On Behalf Of Mark Zelden >>> Sent: Friday, June 25, 2010 3:21 PM >>> To: [email protected] >>> Subject: Re: Delete all members of a PDS that is allocated >>> >>> On Fri, 25 Jun 2010 19:42:27 +0000, Ted MacNEIL >>> <[email protected]> wrote: >>> >>> > >>> >In a z/OS environment, with access methods, online >>> sub-systems, utilities, >>> and the like, less than 5% of the code running on the z/Box >>> is user-written. >>> > >>> >>> Where did you get that statistic from? >>> >>> Mark >>> -- >> >>And does it mean: "5% of the CPU resource used on a z/OS system is consumed >by user written code" or "5% of the number of lines of code for all the >executable programs on a given z/OS system are user written." From context, >I would guess the former. >> > >With all the recovery code written into the OS, subsystems and ISV software, >I would think the latter if anything (but I find that a hard to believe number >also, but could be close to that in a small shop I guess). The system idles >quite nicely at very low utilization, so saying that running production and >development environments at or near 100% (which most cost conscience >shops do) isn't because of "user written code" just because it invokes >access methods, system services, CICS services, SQL or whatever would >be a very misleading statement IMO. > >Mark >-- >Mark Zelden - Zelden Consulting Services - z/OS, OS/390 and MVS >mailto:[email protected] >Mark's MVS Utilities: http://home.flash.net/~mzelden/mvsutil.html >Systems Programming expert at http://expertanswercenter.techtarget.com/ > ---------------------------------------------------------------------- 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
