The XMIT/RECEIVE log (named according to your IKJTSO00 specification) would not necessarily benefit automatically from big blocks. This data set is MODed onto by each XMIT or RECEIVE action. Hence it typically consists of (perhaps very) many small physical blocks regardless of labelled BLKSIZE. The more it's used, the worse performance it gets on READ access.
If you think of it, it's worthwhile now and again EDIT this log file and SAVE it. This rewrites data back to the labelled BLKSIZE. Subsequent READ access will be much improved even at 3120. . . JO.Skip Robinson Southern California Edison Company Electric Dragon Team Paddler SHARE MVS Program Co-Manager 626-302-7535 Office 323-715-0595 Mobile [email protected] From: Jim Mulder <[email protected]> To: [email protected], Date: 07/25/2013 10:46 PM Subject: Re: BLKSIZE=3120 Sent by: IBM Mainframe Discussion List <[email protected]> >> A customer (mildly) complained thatsome of our product allocations still use BLKSIZE=3120. I vaguely remember trying to change all of them to BLKSIZE=0 many years ago (probably before OS/390) and running into some issues with certain IBM utilities. Unfortunately, I can't remember the specifics. In starting to revisit this again, I noticed numerousoccurrences of '3120' in IBM help and documentation. For example, the TSO/E RECEIVE command HELP claims that the log data set must be BLKSIZE=3120: <TSO/E RECEIVE command HELP> LOGDATASET You may specify an alternate data set to be used for the logging of the transmitted data. This data set will be created if it does not exist. The data set should be created with a logical record length of 255, a record format of VB and a blocksize of 3120. ... LOGDSNAME You may specify an alternate data set to be used for the logging of the transmitted data. This data set will be created if it does not exist. The data set should be created with a logical record length of 255, a record format of VB and a blocksize of 3120. </TSO/E RECEIVE command HELP> Is this just outdated help? Or does this restriction still exist? >> I looked at the current TRANSMIT/RECEIVE code, and it still specifies BLKSIZE=3120 when it creates a LOG data set, and still hardcodes BLKSIZE=3120 on its DCB for the log data set. I engaged in battle with the former TSO developers a long time ago (maybe over 20 years ago, maybe even before the advent of System Determined Blocksize). I wanted them to at least remove the BLKSIZE=3120 on the DCB so that if someone (like me) was sensible enough to allocate his own log dataset with an efficient BLKSIZE, TRANSMIT/RECEIVE would cease to override that on the DCB and thus no longer change the BLKSIZE to 3120. Since the stupid DCB specification is still there, exists, I apparently lost that battle. Jim Mulder z/OS System Test IBM Corp. Poughkeepsie, NY ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
