On 25 May 2009 19:16:25 -0700, in bit.listserv.ibm-main you wrote: >> -----Original Message----- >> From: IBM Mainframe Discussion List [mailto:[email protected]] On >> Behalf Of John McKown >> Sent: Monday, May 25, 2009 6:21 PM >> To: [email protected] >> Subject: Re: BLOCK CONTAINS >> >> On Sat, 23 May 2009, Ted MacNEIL wrote: >> >> > >You seem to be agreeing with Steve Thompson that "In the MVS world, >we >> are not device dependant," only insofar as there is only one type of >> device. A weak assertion indeed. >> > >> > Not at all. >> > There are at least two device types -- tape and disk. >> > And, I can convert to either without re-compiling. >> > That is device independent. >> > - >> > Too busy driving to stop for gas! >> >> I agree, except for truly "direct access" data sets. What I have >fought >> with here is the mindset of "I must allocate in CYLINDERS in order to >be >> efficient." I want them to allocate in RECORDS (or millions of >records). >> But, "oh, no! that is not good because I understand what a cylinder >is, >> but I don't know how much space 1 million records requires." Then when >I >> ask them how many records they'll get in that CYLinder allocation, I >get >> the deer in the headlights look. > > >OK, but how is this desire not satisfied via AVGREC allocations? Oh, it >is and you'll claim your applications folks can't handle the concept. >Supply and teach DATACASS with no SPACE at all in the JCL, tell them >it's magic :) Route most if not all allocations to extended format >multivolume in the ACS routines. Use a Big and a not Big set of Storage >groups. Migrate aggressively and route recalls to a pool with less >aggressive free space defined.
If you don't specify round allocation is converted to tracks, not cylinders for non-VSAM. This used to have performance implications before define extent. For VSAM you can get some ugly CA sizes if you specify in records or kilobytes. And it would be valuable for IBM to make ALL FBA type data areas (PDSE's, etc.) readable at NIP and start doing what is necessary to move into the FBA world. VSE can handle FBA. VM can handle FBA. > I've seen hardly any x37 abends in more than a decade. > >> In today's world, if all JCL were allocated in records (not blocks), >(and >> use ROUND if really necesary), then __most__ people wouldn't care one >bit >> about the number of bytes per track or tracks per cylinder or >cylinders >> per volume. They'd care about something more reasonable like number >of >> records or even gigabytes. >> >> And I could have my "beloved" FBA architecture mapped onto standard >SAN >> resident storage. Oh, except for some things like PDSes. PDSes are the >> legacy of the devil, IMO. But the cost to eliminate them would likely >be >> horrendous for things like IPL and NIP. >> >> -- >> Trying to write with a pencil that is dull is pointless. >> >> Maranatha! >> John McKown >> >> ---------------------------------------------------------------------- >> 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 ---------------------------------------------------------------------- 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
