| |
|
RSS Feed - WebSphere MQ Support
|
RSS Feed - Message Broker Support
|
  |
|
| [b]Clustering and load balancing[/b] |
« View previous topic :: View next topic » |
| Author |
Message
|
| cstubbs |
 Posted: Mon Aug 19, 2002 10:31 am Post subject: [b]Clustering and load balancing[/b] |
 |
|
Newbie
Joined: 16 May 2002
Posts: 9
Location: San Antonio
|
I've read a lot of these posts and MQSeries manuals on clustering, and I am still confused on when the round robin algorithm kicks in. Assuming the application specifies NQOO-BIND-NOT-FIXED (might not be exact syntax), when will the messages be sprayed across the clustered qmgrs on the receiving end? Only after a certain workload is achieved? From the onset?
Thanks. [/b] |
|
| Back to top |
|
 |
| kolban |
| Posted: Mon Aug 19, 2002 10:45 am Post subject: |
|
|
Grand Master
Joined: 22 May 2001
Posts: 1072
Location: Fort Worth, TX, USA
|
Each message put to a cluster shared queue when opened with BIND_NOT_FIXED will potentially be put to different queues.
On the first MQPUT, a message will be put to the first queue, on the second MQPUT, the second queue etc until all the cluster queues have been used. Then the next MQPUT will go to the first queue again and so-forth.
There is no workload balancing, only simple round-robin. The queues selected are not affected by the capacity or current processing capabilitity of a machine hosting a queue. |
|
| Back to top |
|
|
| jc_squire |
| Posted: Mon Aug 19, 2002 5:02 pm Post subject: |
|
|
Centurion
Joined: 14 Apr 2002
Posts: 105
Location: New Zealand
|
take a look at page 43 of the clustering manual ....
Quote
The cluster workload exit is called when a cluster queue is opened using the MQOPEN or MQPUT1 call, and when a message is put to a queue using the MQPUT call. The target queue manager selected at MQOPEN time is fixed if MQOO_BIND_ON_OPEN is specified (see MQOPEN). In this case, even if the target queue manager fails, the exit is not run again. In cases where the target queue manager is not fixed, if the target queue manager chosen at the time of an MQPUT call is unavailable, or fails while the message is still on the transmission queue, the exit is called again to select a new target queue manager.
Unquote
Regards
_________________
J C Squire
IBM Certified Specialist - MQSeries |
|
| Back to top |
|
|
| jhalstead |
| Posted: Tue Aug 20, 2002 3:55 am Post subject: |
|
|
Master
Joined: 16 Aug 2001
Posts: 258
Location: London
|
| A small add, if you're doing the puts under syncpoint and put say 10 messages in a single UOW they will all go to the first qmgr... and so on... |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Wed Aug 21, 2002 8:18 pm Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7722
|
Round robining does not kick in if the clustered queue exists locally. In that case all message put will go to that local queue.
QM1, QM2, QM3 and QM4 are in a cluster.
Queue A is defined on locally on both QM1 and QM4, and advertised as a cluster queue.
If an app connects to QM2 or QM3 and starts dumping messages to QueueA,without specifying the bind option, the message will round robin between QM1/QueueA and QM4/QueueA.
If an app connects to QM2 or QM3 and starts dumping messages to QueueA,specifying the bind option, the message will all end up on either QM1/QueueA or QM4/QueueA.
If an app connects to QM1 and starts dumping messages to QueueA, they will ALL go QM1/QueueA, regardless of the bind option.
If an app connects to QM4 and starts dumping messages to QueueA, they will ALL go QM4/QueueA, regardless of the bind option.
See Figure 11 and the accompying text in the Cluster Handbook.
http://publibfp.boulder.ibm.com/epubs/html/csqzah04/csqzah04tfrm.htm
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
|
|
|
| |
|
Page 1 of 1 |
|
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
|
|
|
|
