| |
|
RSS Feed - WebSphere MQ Support
|
RSS Feed - Message Broker Support
|
| MQPut performance |
« View previous topic :: View next topic » |
| Author |
Message
|
| JohnMetcalfe |
 Posted: Mon Jun 20, 2005 3:21 am Post subject: |
 |
|
Apprentice
Joined: 02 Apr 2004
Posts: 40
Location: Edinburgh, Scotland
|
memory use increases initially, then stay pretty stable - as we were execting, as Propagate is supposed to reuse memory.
We have raised a PMR with IBM to get their official response to the problem. |
|
| Back to top |
|
 |
| JohnMetcalfe |
| Posted: Tue Jun 21, 2005 8:53 am Post subject: |
|
|
Apprentice
Joined: 02 Apr 2004
Posts: 40
Location: Edinburgh, Scotland
|
OK, IBM have suggested a fix - see
http://www-1.ibm.com/support/docview.wss?uid=swg1IY63916
As we are on WMQI CSD3, we are seeing if we can get this fix pulled out of the csd9 patch. Looks like it is possible.
However, I'm not convinced this will help. We are not seeing memory use increasing - it is stable.
As another tack, we pulled out our loop/propagate processing into 2 compute nodes.
The first node 'partitions' our messages into 'chunks', which contain a certain number of repeating group structures (i.e rows), and propagates this.
The second compute node receives the propagated message from the first node, spins through the rows in the partioned messages and outputs a message per row (this sound a bit convoluted it is actually very simple).
The performance of this approach seems good - in tests 10,000 messages are being written in 3s to the output Q (compared to 2mins + before). Hence we are applying this 'fudge' to our affected flows and breathing a sign of relief....  |
|
| Back to top |
|
|
| JohnMetcalfe |
| Posted: Tue Jun 21, 2005 8:56 am Post subject: |
|
|
Apprentice
Joined: 02 Apr 2004
Posts: 40
Location: Edinburgh, Scotland
|
Still think this should now go to the WMQI forum - someone there may well be interested. Hey you moderators out there- can you move this thread for me!!!  |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Tue Jun 21, 2005 8:56 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
So by putting in two while/propagate loops, you increased your performance?
Definitely a loop count threshold of some sort...
EDIT: reading the APAR...
| Quote: |
Although parsers were correctly being
re-used on each propagate, this parser was added to a parser
list on every propagate, even though the parser was already
in the list. |
I bet that adding to the end of a 10,000 item list can take a while...
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
|
|
