Ken --
You are basically correct. you can have intervals above 50 as
long as your 4-hour rolling average is below 50. You may
be capped somewhat below 50MSUs, and this depends on the
setup of the LPARs on the machine. To softcap the "weights"
are enforced. If the weight works out to an MSU level below
the 50, then you will be capped there, but a pattern of 'cap
on' 'cap off' 'cap on' will be used to try to average out
the capping to 50.
If the weight is above 50 MSUs then a phantom weight is
calculated to set the cap to 50 MSUs until the 4HRA is below
the defined capacity again.
An interval can exceed use the True Limit of the LPAR (i.e.
all the LPAR's engines at 100%) as long as the 4-hour rolling
average is below the defined capacity.
Al Sherkow
I/S Management Strategies, Ltd.
Consulting Expertise on Capacity Planning, Performance Tuning,
WLC, LPARs, IRD and LCS Software
Seminars on IBM SW Pricing, LPARs, and IRD
Voice: +1 414 332-3062
Web: www.sherkow.com
From: [EMAIL PROTECTED]
Newsgroups: bit.listserv.ibm-main
Subject: Defined Capacity vs. Capping
Date: 19 Jul 2005 11:50:19 -0700
Organization: None
Lines: 39
Message-ID: <[EMAIL PROTECTED]>
NNTP-Posting-Host: mail2.cit.com
Mime-Version: 1.0
Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: 8bit
NNTP-Posting-Date: 19 Jul 2005 18:50:19 GMT
Going to be turning on my new z990 this weekend (figures - now I hear IBM is
about to announce some new boxes soon). Classically we have hard capped our 5
LPARs (contractual restrictions) but to me it seems that using 'defined
capacity' would be beneficial.
Only 'problem' I have is that it is theoretically possible to exceed the
'defined capacity' and have a 'soft cap' kick in which could hurt my
applications.
I am assuming ...
I can use 'defined capacity' even though I am NOT parallel sysplex (I am basic
sysplex - single frame no CF). I know I can't do IRD because of this.
If I 'hard cap' at 50MSU worth I will never go beyond 50MSU (within a small
percentage) and if I am running flat out my 4 hour average would be 50MSU.
If I use 'defined capacity' of 50MSU and I have a period of low utilization
preceding a period of high utilization then I could get more than 50MSU at peak
and still remain within 50MSU four hour average.
Question is if the period of high utilization lasts long enough then I get
'soft capped' to somewhere below 50MSU to keep the four hour average at 50MSU.
Depending on my applications I could be strangling the late comer which is
definitely not good. Whether this is worse than strangling the early bird and
having overlap with the late comer is something I can't predict.
Is there a way of limiting (i.e. min and max) while using 'defined capacity'?
How much will 'defined capacity' allow me to exceed the 50MSU temporarily? Are
there any limits? Does it do a little bit at a time?
Thanks in advance to all ...
Ken Porowski
AVP Systems Software
CIT Group
Email: [EMAIL PROTECTED]
----------------------------------------------------------------------
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
