W dniu 2011-04-15 23:49, Gerhard Adam pisze:
- what do you mean by average. I disclose very big secret: CPU is always
100% busy or 0% busy (*). Always - mean in every tact (tick?).
So the shorter period you take for CPU measurement the higher peaks you
will get. If your periods are as short as
I will be out of the office starting 04/15/2011 and will not return until
04/25/2011.
If you need assistance prior to then please contact Ryan Evans at
216-471-2669.
/pre
This communication may contain privileged and/or confidential information. It
is intended solely for the use of the
The depending answer is 100% correct except when its wrong. :-)
You cannot forecast without knowing the past and it is the that will
tell you whether you need peak or average or more probably a mix.
Hypothetically consider a CEC running at 100% busy. However
simplistically this is a 80/20
- what do you mean by average. I disclose very big secret: CPU is always
100% busy or 0% busy (*). Always - mean in every tact (tick?).
Depends on how many CPs you have.
Date: Fri, 15 Apr 2011 23:37:06 +0200
From: r.skoru...@bremultibank.com.pl
Subject: Re: CPU
W dniu 2011-04-16 10:58, J R pisze:
- what do you mean by average. I disclose very big secret: CPU is always
100% busy or 0% busy (*). Always - mean in every tact (tick?).
Depends on how many CPs you have.
Read carefully. (*) will explain you multi-CPU and other simplifications
I made.
--
W dniu 2011-04-16 10:50, Ken Brick pisze:
The depending answer is 100% correct except when its wrong. :-)
You cannot forecast without knowing the past and it is the that will
tell you whether you need peak or average or more probably a mix.
Hypothetically consider a CEC running at 100% busy.
I'm more interested (and concerned about) performance then capacity planning.
S ...
WLM is basically designed to allow you get the box ticking along at (or very
near) 100% without too much angst to anyone.
I've found that if you happen to have white space, customers prefer to
lower the cap
Neal,
If you think it's pertinent, please check some of the old
APAR's ... I seemed to remember some for the TSO OPER command.
Jim Thomas
-Original Message-
From: IBM Mainframe Discussion List [mailto:IBM-MAIN@bama.ua.edu] On Behalf
Of Neal Eckhardt
Sent: Friday, April 15, 2011 10:45
On Sat, 2011-04-16 at 11:13 +0200, R.S. wrote:
snip
Last, but not least: CPU granularity.
You cannot buy 5% CPU more, or 35%, or 41%. You can buy CP, not half of
it. Situation is much better for smaller machines - mix of n-WAY +
subcapacity levels gives you really good granularity. However
If it gets control in the Master Address Space Where does the Routine
Have to get Loaded??
The documentation for RESMGR describes your choices
-- you can have loaded the routine and provide the address
-- you can have the system issue LINK to get to the routine
-- you can provide a PC number
Mike,
There's 30 years of Papers and discussions on how to do Capacity Planning
for MVS.
The answer is far more complex than the simplicity of your question.
Averages or peaks? What about percentiles and omission of outliers?
A quick Google shows there are courses (Guerilla Capacity Planning),
AMEN! Dr. H Pat is still around over at Performance Associates. SHARE is
another good place for real world experience. Some of Ray Wicks(IBM) old
papers are great and Barry Merrill's(PhD) MXG stuff is continually updated.
In a message dated 4/16/2011 1:00:01 P.M. Central Daylight Time,
On Fri, 15 Apr 2011 03:28:27 -0500, Brian Westerman
brian_wester...@syzygyinc.com wrote:
There are a lot of issues that will pop up over the next few releases of
z/OS with TSSO, mostly because of the changes to the system to support
System REXX, and it's possible also that you have run into
John:
From a memory circa 1980(s). A company in downtown in Chicago got thier
throats
cut with upgrades to CPU capacity. I do not remember the upgrade but the
software alone cost them $250,000 this was mostly (IIRC) a computer associates
bill. Another big cost was a DB2 product add on. That
AMEN! Dr. H Pat is still around over at Performance Associates. SHARE is
another good place for real world experience. Some of Ray Wicks(IBM) old
papers are great and Barry Merrill's(PhD) MXG stuff is continually updated.
Don't forget CMG -- www.cmg.org
Sent wirelessly from my BlackBerry
Well, I disagree with such definition. IMHO CPU busy is pattern of
NOPs (No OPeration) and usable instructions. 1% busy simply means that
99% of cycles were filled with NOP, and only 1% of cycles were other
instructione were executed.
You can disagree with it all you like, but a job that is
W dniu 2011-04-17 01:11, Gerhard Adam pisze:
Well, I disagree with such definition. IMHO CPU busy is pattern of
NOPs (No OPeration) and usable instructions. 1% busy simply means that
99% of cycles were filled with NOP, and only 1% of cycles were other
instructione were executed.
You can
You assumed queue existence. Good assumption for z/OS and most other
systems, but has no meaning from CPU perspective.
A system without a queue has no delays. At that point the only improvement
possible would be a faster architecture.
Again, there are systems, where there are no such thing
On Fri, 15 Apr 2011 16:17:34 -0400 michealbutz michealb...@optonline.net
wrote:
:Would anyone know if installing a Address Space Terminating routine via
RESMGR works Under
:all circumstances
:Meaning normal Termination
Just be aware that it is when the address space terminates. A batch job
I guess if you load the routine it should be in CSA
-Original Message-
From: IBM Mainframe Discussion List [mailto:IBM-MAIN@bama.ua.edu] On Behalf Of
Peter
Relson
Sent: Saturday, April 16, 2011 9:15 AM
To: IBM-MAIN@bama.ua.edu
Subject: Re: RESMGR for terminating Address Space
If it gets
20 matches
Mail list logo