Hi,
No, I have NOT retired! That was funny. Anyway, here is my VMRM
presentation. Also note that there is a COPY file, IRMCONS COPY, that has
some constants defined that can actually be tailored for your system. No
need to recompile anything, the IRMSERV EXEC will pick up the changes when
it runs. VMRM will use these in the algorithms (for example, it uses 5% if
a guest is within this value of their goal. You can change this and
others.) Paul, if you have any questions, please send me an email. I
don't always monitor the Listserv so I may not see questions there. (Maybe
I should check it more often, since it looks like a good rumor mill
website! :-) )
Regards, (See attached file: V62.ppt)
Chris
z/VM Development - Dept. G32/250-2
1701 North St., Endicott, NY 13760
Tieline 620-3434 (607)429-3434
[EMAIL PROTECTED]
Bill Munson
<[EMAIL PROTECTED]
it.state.nj.us> To
The IBM z/VM Operating System
11/20/2007 09:06 <[email protected]>
AM cc
Subject
Re: VMRM?
Paul,
I went to our web site (MVMUA) and this presentation was NOT one of the
ones that we have available for download, sorry.
I the went to www.vm.ibm.com and looked on the developers pages and
unfortunately Chris does NOT have her presentations available for
download either.
Hopefully she will see the postings on this subject and provide YOU a
copy, or it might be on a SHARE proceedings CD that someone might have.
good luck
Bill Munson
VM System Programmer
Office of Information Technology
State of New Jersey
(609) 984-4065
President MVMUA
http://www.marist.edu/~mvmua
Feller, Paul wrote:
> Where could I find the presentation you are talking about?
>
> Thanks..
>
> Paul
>
> -----Original Message-----
> From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED] On
> Behalf Of David Boyes
> Sent: Monday, November 19, 2007 9:20 AM
> To: [email protected]
> Subject: Re: VMRM?
>
>> How many out there have implemented the VM Resource Manager?
>
> Here's one.
>
>> Happy with
>> it? Any tips or gotchas?
>
> It works as specified, adjusting both up and down. It's just a little
> limited.
>
> IMHO, VMRM seems half-finished -- it's not z/OS WLM by any stretch, and
> it's missing some pieces. The CPU goal processing works, but it's fairly
> inflexible, and it's not very clear how you could dynamically adjust
> target workload groups w/o a restart of VMRM (which resets the goal
> targets). The I/O goal processing seems fairly ineffectual, but maybe I
> just don't have the right workloads for it to be effective.
>
> If you're going to enable VMRM, be sure to get a copy of Chris Casey's
> presentation on the topic. The algorithm used to adjust priorities is
> really, REALLY non-intuitive -- Chris's presentation explains it much
> better than the docs do. Steve Wilkins has given it the last few times
> I'm aware of an opportunity.
>
> I think Chris may have retired recently. If so, I'm not sure who has
> taken VMRM up as maintainer. It's also not really clear what role it
> plays in the bigger IBM Director plan -- it has some role in that, but I
> don't understand what it is.
>
