Acutally, if the jobs are running consecutively, at pretty much the same time, PAV will not be that great a help. The PAV is designed for multiple accesses on different extents. Since all the jobs will be acessing the same extent range, only one UCB will probably be used.
I like the idea of using HiperBatch the best. Work up a job to load the dataset into hiperbatch buffers, then let the 35 jobs go at it. When done, just unload the hiperbatch stuff. On Mon, 1 May 2006 17:01:19 +0200, Vernooy, C.P. - SPLXM <[EMAIL PROTECTED]> wrote: >"Bob Shannon" <[EMAIL PROTECTED]> wrote in message news:<[EMAIL PROTECTED]>... >> >"How many jobs can access a DISK DSN with a DISP=SHR before any >> >performance degradation occurs due to access contention? >> >> >"Meaning, We have 1 dataset sitting on disk, we have 35 jobs that need >> >to access this one dataset, how many jobs can run at one time accessing >> >this one dataset before the access creates contention and stops jobs >> >from running?" >> >> >All of the access would be read-only, and AFAIK the dataset is >> non-VSAM. >> >> Sounds like an ideal environment Hiperbatch. It's still supported, >> and this sounds like the exact scenario it was created to address. >> >> >> >> > > >Sounds also like an ideal environment for PAVs. > >Kees. ---------------------------------------------------------------------- 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
