Re: cxx performance
From: jlsue (jefflsxxxz_at_sbcglobal.net)
Date: 07/14/03
- Next message: Bob Koehler: "Re: Cluster satellite fails to boot"
- Previous message: Jack Pea***: "Re: OT: HP's PeeCee customer statisfaction"
- In reply to: David J. Dachtera: "Re: cxx performance"
- Next in thread: David J. Dachtera: "Re: cxx performance"
- Reply: David J. Dachtera: "Re: cxx performance"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Date: Mon, 14 Jul 2003 19:54:32 GMT
On Fri, 04 Jul 2003 11:22:22 -0500, "David J. Dachtera"
<djesys.nospam@fsi.net> wrote:
>
>Actually, its not that complex. WSDEF harks back to the days when memory
>was expensive and tended to be somewhat dear. Now that VMS is more
>frequently found on memory-rich systems, it makes sense to let WSDEF =
>WSQUOTA in such cases. That way, the working set doesn't have to be
>expanded just to do useful work. You can set it to a value most likely
>to be needed. Then, when an image is activated, all the system has to do
>is read it in - no need to expand the working set first.
I totally agree that WSDEF=WSQUOTA is fine for many, many systems now.
But make sure that everyone understands that it's NOT the working set that
expands, necessarily, but the WS List Size. I.e., as Bruce Ellis was fond
of saying:
Q: "How much memory does VMS give your working set when
you login?"
A: "None! You've got to earn it."
What this means is that you WS List Size=WSDEF(=WSQUOTA if you make this
change). However, you don't actually have that memory allocated to you
unless you need it (ask for it).
- Next message: Bob Koehler: "Re: Cluster satellite fails to boot"
- Previous message: Jack Pea***: "Re: OT: HP's PeeCee customer statisfaction"
- In reply to: David J. Dachtera: "Re: cxx performance"
- Next in thread: David J. Dachtera: "Re: cxx performance"
- Reply: David J. Dachtera: "Re: cxx performance"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]