SYS1.MAN allocation question
We've always allocated our SMF files as follows: DEFINE CLUSTER( - CONTROLINTERVALSIZE(26624) - TRACKS(6300) - NAME(SYS1.MAN2) - NONINDEXED - RECORDSIZE(26614,32767) - REUSE - SHAREOPTIONS(2) - SPANNED - SPEED - VOLUME(SYS004) ) I'm just curious since I don't see any optimal settings in any of the manuals for recordsize and cisize is this a good allocation?? Any help would be appreciated. We are at z/OS 1.7. -- 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: SYS1.MAN allocation question
Below is the documented define from the SMF manual. Your Cisize of 26K seems to be optimum for a 3390, see section 2.1.1.2 Using DEFINE to Create SMF Data Sets of z/OS V1R4.0 MVS System Management Facilities (SMF). I'm a bit curious as to why you would change the recommended Recordsize? How many hours of recording does 6300 tracks give you? DEFINE CLUSTER (NAME(SYS1.MANX) + VOLUME(xx) + NONINDEXED + CYLINDERS(nn)+ REUSE+ RECORDSIZE(4086,32767) + SPANNED + SPEED+ CONTROLINTERVALSIZE()+ SHAREOPTIONS(2)) -Original Message- From: George Dranes [mailto:[EMAIL PROTECTED] Sent: Wednesday, November 15, 2006 9:26 AM To: IBM-MAIN@BAMA.UA.EDU Subject: SYS1.MAN allocation question We've always allocated our SMF files as follows: DEFINE CLUSTER( - CONTROLINTERVALSIZE(26624) - TRACKS(6300) - NAME(SYS1.MAN2) - NONINDEXED - RECORDSIZE(26614,32767) - REUSE - SHAREOPTIONS(2) - SPANNED - SPEED - VOLUME(SYS004) ) I'm just curious since I don't see any optimal settings in any of the manuals for recordsize and cisize is this a good allocation?? Any help would be appreciated. We are at z/OS 1.7. -- 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: SYS1.MAN allocation question
On Wed, 15 Nov 2006 08:25:46 -0600, George Dranes [EMAIL PROTECTED] wrote: We've always allocated our SMF files as follows: DEFINE CLUSTER( - CONTROLINTERVALSIZE(26624) - TRACKS(6300) - NAME(SYS1.MAN2) - NONINDEXED - RECORDSIZE(26614,32767) - REUSE - SHAREOPTIONS(2) - SPANNED - SPEED - VOLUME(SYS004) ) I'm just curious since I don't see any optimal settings in any of the manuals for recordsize and cisize is this a good allocation?? Any help would be appreciated. We are at z/OS 1.7. That looks like what I've been using since MVS/ESA 4.3 (for a 3390). The only thing different I have is: BUFFERSPACE(106496) - I'm not sure what IBM manuals or Redbooks say (and if you checked the right places) but Cheryl Watson has probably hit on this more than once over the years. I'm sure it gets talked about on MXG-L also and at CMG. Regards, Mark -- Mark Zelden Sr. Software and Systems Architect - z/OS Team Lead Zurich North America / Farmers Insurance Group - GITO mailto:[EMAIL PROTECTED] z/OS and OS390 expert at http://searchDataCenter.com/ateExperts/ Systems Programming expert at http://expertanswercenter.techtarget.com/ Mark's MVS Utilities: http://home.flash.net/~mzelden/mvsutil.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: SYS1.MAN allocation question
On Wed, 15 Nov 2006 08:25:46 -0600, George Dranes [EMAIL PROTECTED] wrote: We've always allocated our SMF files as follows: DEFINE CLUSTER( - CONTROLINTERVALSIZE(26624) - TRACKS(6300) - NAME(SYS1.MAN2) - NONINDEXED - RECORDSIZE(26614,32767) - REUSE - SHAREOPTIONS(2) - SPANNED - SPEED - VOLUME(SYS004) ) Compatibly with Mark Zelden's BUFFERSPACE(106496) recommendation and your own CONTROLINTERVALSIZE(26624) value, I would change RECORDSIZE(26614,32767) which may be an old typo, to RECORDSIZE(26624,32767) John Gilmore Ashland, MA 01721-1817 USA _ Use your PC to make calls at very low rates https://voiceoam.pcs.v2s.live.com/partnerredirect.aspx -- 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: SYS1.MAN allocation question
There's been some disagreement over the specification of CISIZE for the SMF datasets. A CISIZE of 26624 gives you 52K per track on a 3390 (2 CIs) ... slightly better performance. A CISIZE of 18432 will give you 54K per track on a 3390 (3 CIs) ... sightly better space utilization. Make your own choice. And, as always, YMMV. The BUFSP parameter should typically be 5 times the CISIZE. The average (first number in) RECSZ value pretty much doesn't matter (in this case); the maximum (second number in) RECSZ value MUST be 32767. -- 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: SYS1.MAN allocation question
On Wed, 15 Nov 2006 16:39:58 +, john gilmore [EMAIL PROTECTED] wrote: snip recommendation and your own CONTROLINTERVALSIZE(26624) value, I would change RECORDSIZE(26614,32767) which may be an old typo, to RECORDSIZE(26624,32767) Not a typo today, but I'm not enough of a VSAM expert to say if there is any real reason why the avg recordsize is less than the CISIZE specification. I'm pretty sure I originally got those numbers from Cheryl Watson or a CMG paper. I also found those same specifications in this Redbook: System/390 MVS Parallel Sysplex Continuous Availability SE Guide SG24-4503-00 Mark -- Mark Zelden Sr. Software and Systems Architect - z/OS Team Lead Zurich North America / Farmers Insurance Group - GITO mailto:[EMAIL PROTECTED] z/OS and OS390 expert at http://searchDataCenter.com/ateExperts/ Systems Programming expert at http://expertanswercenter.techtarget.com/ Mark's MVS Utilities: http://home.flash.net/~mzelden/mvsutil.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: SYS1.MAN allocation question
Mark Zelden wrote: On Wed, 15 Nov 2006 16:39:58 +, john gilmore [EMAIL PROTECTED] wrote: snip recommendation and your own CONTROLINTERVALSIZE(26624) value, I would change RECORDSIZE(26614,32767) which may be an old typo, to RECORDSIZE(26624,32767) Not a typo today, but I'm not enough of a VSAM expert to say if there is any real reason why the avg recordsize is less than the CISIZE specification. CIDF+2RDFs ? AFAIR CIDF is 4 bytes, RDFs are 3B each. Sum is 10 bytes. BTW: This is SPANNED file. For non-spanned clusters obviously it's common to have avg recordsize less than CISZ. For SPANNED it's not so obvious, but still avg can be a value outside of set of allowable CI sizes. 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: SYS1.MAN allocation question
So what I take from this discussion is that what I currently have is the recommended settings? I should add BUFFERSPACE(106496) to my define. Can I change the bufferspace value to each MAN file individually or is this one of those changes that require all MAN files be redefined? -Original Message- From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of Mark Zelden Sent: Wednesday, November 15, 2006 11:13 AM To: IBM-MAIN@BAMA.UA.EDU Subject: Re: SYS1.MAN allocation question On Wed, 15 Nov 2006 16:39:58 +, john gilmore [EMAIL PROTECTED] wrote: snip recommendation and your own CONTROLINTERVALSIZE(26624) value, I would change RECORDSIZE(26614,32767) which may be an old typo, to RECORDSIZE(26624,32767) Not a typo today, but I'm not enough of a VSAM expert to say if there is any real reason why the avg recordsize is less than the CISIZE specification. I'm pretty sure I originally got those numbers from Cheryl Watson or a CMG paper. I also found those same specifications in this Redbook: System/390 MVS Parallel Sysplex Continuous Availability SE Guide SG24-4503-00 Mark -- Mark Zelden Sr. Software and Systems Architect - z/OS Team Lead Zurich North America / Farmers Insurance Group - GITO mailto:[EMAIL PROTECTED] z/OS and OS390 expert at http://searchDataCenter.com/ateExperts/ Systems Programming expert at http://expertanswercenter.techtarget.com/ Mark's MVS Utilities: http://home.flash.net/~mzelden/mvsutil.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 -- 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: SYS1.MAN allocation question
Can I change the bufferspace value to each MAN file individually or is this one of those changes that require all MAN files be redefined? You may try to run IDCAMS: ALTER SYS1.MAN1.D BUFSP(110592) ALTER SYS1.MAN3.D BUFSP(110592) ALTER SYS1.MAN4.D BUFSP(110592) ALTER SYS1.MAN5.D BUFSP(110592) ALTER SYS1.MAN6.D BUFSP(110592) ALTER SYS1.MAN7.D BUFSP(110592) ALTER SYS1.MAN8.D BUFSP(110592) ALTER SYS1.MAN9.D BUFSP(110592) (this is from my old JCL) against all your active SMF datasets. This may work and should not hurt anything. If you are adding a new SMF dataset to the existing ones - you cannot change the CISIZE. I think it is even worse, to change CISIZE you need to re-IPL with a redefined set of SMF datasets (SET SMF inactive/active will not work). -- 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
