+1 for Martins blog post, which is excellent.
Having said that, the whole zIIP concept is baroque from root to branch.
Why couldn't IBM have come up with something much simpler?
On 23/01/2014 8:29 PM, Scott Chapman wrote:
I think that's a better overview of the issue Martin, because you bring out the
important fact that z/OS doesn't forget about how to prioritize work just
because it's on a specialty engine. It's probably would also be a similarly
good rule of thumb that you shouldn't let your GPs get over 50% busy with the
most important system-related work. And having more dispatchable CPUs than less
has been better in my experience.
One small point though relative to your blog post: each DDF transaction
actually now creates two enclaves: an indepedent enclave and a work-dependent
enclave. It seems that the work-dependent enclave runs only on the GCP and does
40% of the work, while the independent enclave runs 100% on the zIIP and does
60% of the work. But it does appear from the doc and data that work dependent
enclave CPU time is captured in the indepedent enclave CPU time buckets, which
makes sense.
Scott
On Thu, 23 Jan 2014 09:46:18 +0000, Martin Packer <[email protected]>
wrote:
I'm also talking about this. In the final throes of having a presentation
I aim to put on Slideshare (and give at any suitable venue).
The message is important: DB2 V10 changes the way we view zIIP capacity.
The other message is important: zIIP Capacity Planning IS very feasible.
And at the risk of annoying people with blatant self promotion :-) see:
https://www.ibm.com/developerworks/community/blogs/MartinPacker/entry/ziip_address_space_instrumentation?lang=en
Cheers, Martin
Martin Packer,
zChampion, Principal Systems Investigator,
Worldwide Banking Center of Excellence, IBM
+44-7802-245-584
email: [email protected]
Twitter / Facebook IDs: MartinPacker
Blog:
https://www.ibm.com/developerworks/mydeveloperworks/blogs/MartinPacker
From: Norbert Friemel <[email protected]>
To: [email protected]
Date: 23/01/2014 09:01
Subject: Re: Resistance to Java.
Sent by: IBM Mainframe Discussion List <[email protected]>
On Thu, 23 Jan 2014 15:59:23 +0800, David Crayford wrote:
On 22/01/2014 11:53 PM, Skip Robinson wrote:
IBM is currently warning customers that over-using ZIIPs may lead
to serious performance problems because of the way z/OS manages them
vs.
the way it manages general purpose CPs.
Can you supply a link wrt that information?
Google found:
https://www.ibm.com/developerworks/community/blogs/22586cb0-8817-4d2c-ae74-0ddcc2a409bc/entry/december_17_2012_6_07_am3?lang=en
http://ibmdatamag.com/2013/04/performance-and-capacity-considerations-for-ziip-engines/
Norbert Friemel
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number
741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
