Thomas Berg wrote:
>I'm using TSO RECEIVE to read outputs from JES SPOOL.
On what z/OS level are you?
>The outputs is created by allocating and writing as:
>Rc = BPXWDYN('ALLOC LOGROUTT SYSOUT(8) SPIN(UNALLOC) DEST(systarget.userid)
>RECFM(V,B) LRECL(varies) BLKSIZE(32756) REUSE')
>'EXECIO' s.0 'DISKW LOGROUTT (STEM S. FINIS'
>The problem is the performance. There is many files with mostly one record.
>Receive takes from 1/10 to 1 second elapsed time to get one file.
Ouch. And you probably have gazillions of them?
>I have tried the REXX/ISFACT interface to SDSF but it was only a minor, if
>any, improvement.
I would also go that route. But in the first place I would like the jobowner to
write the outputs in the first place to datasets. Say for example changing that
ALLOC statements somehow.
Possible alternatives which I have NOT tested out for your scenario: Offloader?
External writer, probably RYO one?
I think there are Assembler (better performance) examples on the CBTTAPE.
Anyone out there who can remember in what file(s) it was?
We have a commercial spool offloader we use to extract the job outputs and
place it in datasets which can be browsed in an ISPF screen. That offloader
frees up our spool space by extracting job outputs 24 hours per day and keep it
for x days as required by our users.
>Any suggestion for a better performance?
Not much. There is a lot of overhead for just one record as found PER job. JES2
and your interface to open it from both sides. Then the actual transfer of that
line and then close on both sides. Then JES2 has to update its checkpoints and
do some spooloverhead. Then your interface has to unalloc it and catalog it.
And all of this - if your JES2 is really busy, you will have to wait your turn.
Perhaps little WLM tuning can help?
Groete / Greetings
Elardus Engelbrecht
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
