Hi Ron: I dumped a CA split on our system and the "old" CIs that were moved to the new CA were cleared to low values in the original CA formatted with the appropriate CIDF. So,I believe that these empty CIs will occupy space in the disk cache as they are written by VSAM. Based on the issue you raised that the hardware vendor recognizes the sequential bias in the command, do the pre-formatted CIs get read into the cache as the hardware tries to anticipate the data required when reading the file sequentially? Regards, Gene In a message dated 2/21/2008 3:18:10 A.M. SA Western Standard Time, [EMAIL PROTECTED] writes:
Gene, I have not heard of a CA being "cleared out." My old notes say that all CI with a key higher than the mid-point sequence set record are written to the new CA. The space in the old CA is now marked free - it is not formatted. If you are using CA of 1 CYL then it is unlikely to upset the pre-fetch algorithm in most storage. The sequence set within a CA are in order and with default buffering (BUFND=2) or SEQ bias (half CYL) you will get sequential cache hits. That assumes that your disk vendor recognizes the Seq bit. This is not the case for all vendors. A heavily fragmented KSDS may not have pre-fetch detected with a large BUFND because it jumps out of the monitored area before enough seq IO takes place. It is likely that the CI split that instigated the CA split will be a DFW hit because the old tracks are still in cache - but that's a nit. Ron BTW my notes make a post of mine about the duration of CA splits wrong. It is only halve IO cost I estimated. I should have check first. Where's my "Ronald K. Ferguson" bible when I need it? > -----Original Message----- > From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On > Behalf Of Gene Hudders > Sent: Wednesday, February 20, 2008 1:02 PM > To: IBM-MAIN@BAMA.UA.EDU > Subject: [IBM-MAIN] Questions Regarding Disk Cache > > Hi: > > These were a few questions posted on another topic in CICS-L for which > no > reply was received. The topic was CI/CA splits and I asked the > following > questions, which were embedded in the response: > > 1) Two of the processes of completing a CA split is to format the new > CA and > clear out the moved data CIs from the original CA. Do the "empty" CIs > used > to clear or format the CAs occupy space in the disk cache? > 2) When reading a file sequentially that has CA splits caused by direct > insertions (e.g., 50% free space), are the empty CIs that were created > as a > result of the CA splits read into the disk cache by the hardware's > read-ahead > mechanism once it determines that the file is being processed > sequentially? > > Regards, > Gene > > > > > > > **************Ideas to please picky eaters. Watch video on AOL Living. > (http://living.aol.com/video/how-to-please-your-picky-eater/rachel- > campos-duffy/ > 2050827?NCID=aolcmp00300000002598) > > ---------------------------------------------------------------------- > 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 **************Ideas to please picky eaters. Watch video on AOL Living. (http://living.aol.com/video/how-to-please-your-picky-eater/rachel-campos-duffy/ 2050827?NCID=aolcmp00300000002598) ---------------------------------------------------------------------- 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
