Hi Gary,
I have a very similar setup though my highest availability/performance
critical systems are on a separate CEC.
#2 OS tasks darn well better be where IBM told you to put them regardless of
which system they are on. If you are in the same Sysplex and z/OS on LPAR A
fouls up it can hurt LPAR B. In a related thought make sure you have a good
Sysplex Failure Management (SFM) Policy.
#0 I have not seen the needs for any more than one level of importance
separation. I also like some others here run almost all my test batch as
discretionary which is how I soak up cycles that would otherwise go to
waste. We don't run any user work on test LPAR's at IMP 1 (SYSTEM and
SYSSTC tasks are present). Remember your job is to find someone to kick
around and make sure WLM will know when PROD is not meeting it's goals.
An old MVS systems manager once told me 'put test batch jobs somewhere below
whale excrement'. Still good advice IMHO.
I do not set minimum or maximum weights and have not seen a problem from
this yet.
So far any issues I have seen with IRD were not IRD problems but WLM policy
problems. If you don't like what IRD does you probably need to look at your
WLM policy.
Best Regards,
Sam Knutson, GEICO
Performance and Availability Management
mailto:[EMAIL PROTECTED]
(office) 301.986.3574
"Think big, act bold, start simple, grow fast..."
-----Original Message-----
In setting up one CEC that has multiple LPARS, some TEST and some PROD, with
one combined WLM Policy, all systems in the same plex...
How much do the WLM Policies have to differ? TEST has to have lower goals
than PROD, obviously, with IRD and CPU Vary enabled. But by how much?
Does anyone have direct experience with this that can talk a little bit
about it? Specifically, I guess I'm really asking:
With IRD weight management (and low minimum weights, no max weight), and CPU
Vary enabled (and ALL LPs put in Initial so ANY LPAR can get ANY number of
LPs):
1. Should TEST LPAR work be one, or two importance levels below PROD? (or
more?!?)
2. Should the operating system tasks still run at SYSSTC on TEST (i.e. at
the same high service class level that the same named tasks run on PROD)?
I'd like to run TEST uncapped, and let it "run away with the box" when
utilities are running and PROD is IDLE, but ensure that PROD gets priority
whenever it needs it.
Best regards,
Gary Diehl
<>
====================
This email/fax message is for the sole use of the intended recipient(s) and
may contain confidential and privileged information. Any unauthorized
review, use, disclosure or distribution of this email/fax is prohibited. If
you are not the intended recipient, please destroy all paper and electronic
copies of the original message.
----------------------------------------------------------------------
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
