Author |
Message
|
Allan Mackay |
Posted: Wed Oct 23, 2013 1:25 am Post subject: MQ7.1 page set usage versus MQ6 |
|
|
Apprentice
Joined: 08 Jul 2010 Posts: 29
|
Hi,
I have a bit of a puzzle I can't make sense of.
Back last May 2013, we ran a batch job that loads a queue, on a zOS queue manager, with a large number (>4m) of short (20 byte) messages. The queue in question is defined on a "large volume" page set (1.04m pages). The 4.1m messages fitted comfortably (I couldn't tell you now how comfortably).
Yesterday, we ran the very same job and it filled the page set after circa 833k messages. Admittedly, the page set was already at 19% used. But 833k messages using the remaining 81% as opposed to 4.1m fitting comfortably?
Looking at the the 6.0 and the 7.1 manuals around page set usage, nothing appears to have changed. According to rough calculations, we should have been able to fit 10 messages per page. Thus, roughly again, 4.1m should occupy about 40% of the page set. It now looks like, on the face of it, each short message is now occupying its own page...
Am I being naive? Or have I missed something blindingly obvious between the two versions of MQ? Or in the configuration of the new version of MQ?
Any hints, tips or derision coming this way will be gratefully received as I am at a loss to explain this at the moment...
-Allan. |
|
Back to top |
|
 |
elkinsc |
Posted: Wed Oct 23, 2013 2:23 am Post subject: Please look at Performance Report MP1G |
|
|
 Centurion
Joined: 29 Dec 2004 Posts: 138 Location: Indy
|
Storage of small messages changed between V6 and V7.0.1. The Performance report is the only place this behavior was documented. |
|
Back to top |
|
 |
Allan Mackay |
Posted: Wed Oct 23, 2013 2:36 am Post subject: |
|
|
Apprentice
Joined: 08 Jul 2010 Posts: 29
|
Ooh! Thanks elkinsc, just what I needed  |
|
Back to top |
|
 |
elkinsc |
Posted: Wed Oct 23, 2013 2:43 am Post subject: You are welcome |
|
|
 Centurion
Joined: 29 Dec 2004 Posts: 138 Location: Indy
|
This information is difficult to find. |
|
Back to top |
|
 |
PeterPotkay |
Posted: Wed Oct 23, 2013 4:11 am Post subject: Re: MQ7.1 page set usage versus MQ6 |
|
|
 Poobah
Joined: 15 May 2001 Posts: 7722
|
|
Back to top |
|
 |
Allan Mackay |
Posted: Wed Oct 23, 2013 6:37 am Post subject: |
|
|
Apprentice
Joined: 08 Jul 2010 Posts: 29
|
Thanks PeterPotkay. Part of the issue, I think, is that we jumped from 6.0 directly to 7.1... so did not look at infocenter for 7.0...  |
|
Back to top |
|
 |
PeterPotkay |
Posted: Wed Oct 23, 2013 6:48 am Post subject: |
|
|
 Poobah
Joined: 15 May 2001 Posts: 7722
|
|
Back to top |
|
 |
elkinsc |
Posted: Wed Oct 23, 2013 7:18 am Post subject: Originally |
|
|
 Centurion
Joined: 29 Dec 2004 Posts: 138 Location: Indy
|
It was not even in the v7 doc at all, until several people complained. it never did get highlighted as a change between releases.
Still the best place to get the information on the way it works is from that Performance Report. Tony Sharkey does a really good job on explaining the behavior and the impact of turning this feature off, something some customers have been forced to so to accommodate regular very deep queues of small messages.
Have a great evening ! |
|
Back to top |
|
 |
PeterPotkay |
Posted: Wed Feb 05, 2014 8:21 am Post subject: |
|
|
 Poobah
Joined: 15 May 2001 Posts: 7722
|
I received an email from the IBM documentation team this morning saying that they accepted my recomendation to have this info included in the 7.1 and 7.5 InfoCenters, like it is already in the 7.0 Info Center.
Quote: |
Writer's Response:
Thank you for your comment. This item is being dealt with using an internal work item. Your suggested amendments will be seen in future refreshes of the WebSphere MQ information centers.
|
_________________ Peter Potkay
Keep Calm and MQ On |
|
Back to top |
|
 |
elkinsc |
Posted: Sat Feb 08, 2014 6:00 am Post subject: |
|
|
 Centurion
Joined: 29 Dec 2004 Posts: 138 Location: Indy
|
Thank you Peter for sending the reader comment on that. It is always taken more seriously when it comes from a customer instead of an internal source.  |
|
Back to top |
|
 |
|