Size isn't the issue, it's being able to dump them quicker than they
fill up.
Indeed. Doesn't SMF support use of the logger? Does anyone have
experience using the logger for SMF?
SMF doesn't write to logger (out of the box). We do write SMF to logger,
but
I wrote the code to do it. Pretty interesting project really.
Getting back to the subject of MAN datasets, we had an issue on one system
in
a previous century, an application decided to do excessive DB2 stuff. We
got
up to 12 or 16 MAN datasets and still had issues.
The primary thing we did to resolve the problem was to improve service to
the
dump process. Obviously, the dump process needs to run at least as well as
the creating process. Then we changed it so that each dump process created
a unique dataset name, so multiple dump processes could run in parallel.
That
required a little rexx code to query the catalog and build the JCL for the
daily job. But now we can run parallel dump processes and we can run the
dump
processes in parallel with the daily accumulation job. We run three MAN
datasets,
70 - 75 cylinders each. Only time we use the third MAN is when the onlines
come
down and batch starts up. This technique has worked for several processor
generations on 38 LPARs (down from 64).
Lance
--
Shmuel (Seymour J.) Metz, SysProg and JOAT
----------------------------------------------------------------------
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
