Question about dataset allocations

This forum provides the support of Dezhi Mainframe systems. Please post your questions about logon, usage of our mainframe environment.

Moderators: sysprog, prino, sfan, steve-myers, Tim001

Question about dataset allocations

Postby axdmlm » Tue 15 Aug 2017, 04:22

When I logon to TSO I see an admonition about proper blocksizes for the system.
When I allocate data sets via ISPF the defaults differ from those specified in the logon message the logon messages:
RECFM=FB,LRECL=80 datasets with a BLKSIZE other than 27920
RECFM=VB,LRECL=ANY datasets with a BLKSIZE other than 27998

ISPF:
Record format . . . . FB
Record length . . . . 80
Block size . . . . . 32000
Are the ISPF defaults legal?
axdmlm
 
Posts: 7
Joined: Sun 13 Aug 2017, 17:28
Location: Arroyo Grande, California USA

Re: Question about dataset allocations

Postby steve-myers » Tue 15 Aug 2017, 07:06

Yes, but they are not optimal. If you are using ISPF 3.2 to allocate data sets, leave the BLKSIZE area blank and ISPF (well, the system, not ISPF) will create a more "optimal" BLKSIZE.

In JCL, if you are allocating a sequential data set, specify DCB=(RECFM=FB,LRECL=80,DSORG=PS) to achieve a similar effect

BLKSIZE=32000 in RECFM=FB,LRECL=80,BLKSIZE=32000 is nearly as bad as BLKSIZE=80 or BLKSIZE=800, which gets the admins in a dither and often prompts them to "ban" the userid as being that of a trainee when they see them. They tolerate BLKSIZE=32760 (the default) for load module data sets because the Binder rarely, if ever, writes real records that big.
steve-myers
 
Posts: 466
Joined: Tue 04 May 2010, 15:43

Re: Question about dataset allocations

Postby prino » Tue 15 Aug 2017, 14:25

axdmlm wrote:When I logon to TSO I see an admonition about proper blocksizes for the system.

Which really shows you're a real z/OS user (which I already guessed from your name, it's very familiar, but I don't remember from where). Just use the system, and if you screw up occasionally, don't worry (too) much.
Robert AH Prins
robert.ah.prins @ the.17+Gb.Google thingy
Some programming here :mrgreen:
prino
 
Posts: 491
Joined: Sat 06 Jun 2009, 21:41
Location: Oostende, Belgium

Re: Question about dataset allocations

Postby axdmlm » Wed 16 Aug 2017, 00:18

I do not wish to belabor the point, but the logon messages suggested using the DCOPYD command from option 3.4 to create a Library with a more optimal blocksize. When I did that, option I for that Library showed the blocksize as
AXDMLM.MACLIB.S00.COPY
Record length . . . : 80
Block size . . . . : 32720
1st extent tracks . : 63
Secondary tracks . : 63
Data set name type : LIBRARY

I also tried to allocte a Library by clearing the blocksize field from option 3.2.
That resulted the following:
AXDMLM.MACLIB4
Record format . . . : FB
Record length . . . : 80
Block size . . . . : 32720
1st extent tracks . : 63
Secondary tracks . : 63
Data set name type : LIBRARY

I certainly know how to correct these matters, but I thought you might like to know my experience.
axdmlm
 
Posts: 7
Joined: Sun 13 Aug 2017, 17:28
Location: Arroyo Grande, California USA

Re: Question about dataset allocations

Postby steve-myers » Wed 16 Aug 2017, 03:43

For PDSE type data sets, the BLKSIZE is essentially immaterial. The actual data is mapped over real 4K records.
steve-myers
 
Posts: 466
Joined: Tue 04 May 2010, 15:43

Re: Question about dataset allocations

Postby axdmlm » Wed 16 Aug 2017, 04:23

I had forgotten that.
Thanks for the reminder.
I suppose that is why the default is Library.
axdmlm
 
Posts: 7
Joined: Sun 13 Aug 2017, 17:28
Location: Arroyo Grande, California USA

Re: Question about dataset allocations

Postby steve-myers » Wed 16 Aug 2017, 04:36

axdmlm wrote:... I suppose that is why the default is Library.

Actually, no. If you use ISPF 3.2 to obtain a prototype the type of the prototype is retained for the real allocation.

Personally, I rarely use PDSE unless I need its data management capabilities. It almost always requires more real disk space than PDS, and requires more CPU to process.
steve-myers
 
Posts: 466
Joined: Tue 04 May 2010, 15:43

Re: Question about dataset allocations

Postby prino » Wed 16 Aug 2017, 07:25

I've changed DCOPY(D) to actually reblock datasets to half-track blocking, they're in 'sysfan.user.rexxlib', which has RACF update access so if you feel like sharing some of your code, feel free to do so there. I've also put both under the provisions of the GPL V3.
Robert AH Prins
robert.ah.prins @ the.17+Gb.Google thingy
Some programming here :mrgreen:
prino
 
Posts: 491
Joined: Sat 06 Jun 2009, 21:41
Location: Oostende, Belgium


Return to Dezhi systems: Mainframe

Who is online

Users browsing this forum: No registered users and 1 guest

cron