In <[EMAIL PROTECTED]>, on 07/26/2005
at 09:32 PM, Ron and Jenny Hawkins <[EMAIL PROTECTED]> said:
>However, with PLPA overflowing into Common the majority of PLPA
>page-ins will come from the Common Page Dataset, not PLPA.
Correct, but the question is whether they will be frequent enough to
be relevant for performance.
>A dedicated PLPA means there are "always" two exposures for PLPA
>Page-ins, and "always" two exposures for Common page-in and
>page-out.
Sure, but with the PLPA activity levels that I've seen, there would be
no measurable difference in performance. Certainly not enough to
justify your "silly" or even to repay the effort of periodically
resizing PLPA.
>As for normal activity on the PLPA dataset, well have you considered
>that the activity on a one Cylinder PLPA dataset does not represent
>the page-in rate for PLPA pages.
Have you ever considered that the measurements are not based on
dataset level I/O counts?
>So tell me, if I am building a new system and allocating new PLPA,
>Common, and Locals, why wouldn't I simply allocate a larger PLPA
>without overflow this time around (with PAV).
It's not my dog; you will do what you will do. My concern is in your
claiming that everyone else is being silly for not copying you.
>That would be two JCL changes:
What will you do when service expands the PLPA? Have you taken that
into account?
>Net result is improved throughput in the event of heavy Page-in
>activity for PLPA and Common pages,
I'll worry about what to feed the unicorn when I see one.
>Change it or not is a pointless debate, as YMMV. Which configuration
>should be used to provide the maximum potential page-in rate is what
>I am discussing.
While I am discussing how to configure it on a real system with a real
load.
--
Shmuel (Seymour J.) Metz, SysProg and JOAT
ISO position; see <http://patriot.net/~shmuel/resume/brief.html>
We don't care. We don't have to care, we're Congress.
(S877: The Shut up and Eat Your spam act of 2003)
----------------------------------------------------------------------
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
