Author |
Message
|
csmith28 |
Posted: Tue Mar 14, 2006 7:58 pm Post subject: Weblogic WLI using JMS to browse then get..... |
|
|
 Grand Master
Joined: 15 Jul 2003 Posts: 1196 Location: Arizona
|
I've got this application running on WebLogic using WLI, three application instancs using JMS to connect to the MQManager.
The Developers are insisting that they need 40 Channels / application instance to listen to the Queue because the application doesn't actually GET the message until after they get a response from another application.
That is to say, the application does a BROWSE of a message but the message itself stays on the request Queue until the application gets a reply or a timeout from from this other application. They are telling me that each SVRCONN Channel is then devoted to that thread and unable to do any other work until the thread gets a response or a timeout. Then the Channel GET's the message and can move on.
In other words they say they need to have 120 Channels listening to the queue becuase the application needs to be able to handle as many as 120 concurrent transactions and each of these transaction have to have a devoted SVRCONN Channel instance.
This is contrary to all I understand about SVRCONN Channel behavior.
Is this or could this be correct? I don't have access to the code. The applications are running on Linux/WebLogic. The MQManager is 5.3.0.6 running on AIX. _________________ Yes, I am an agent of Satan but my duties are largely ceremonial. |
|
Back to top |
|
 |
PeterPotkay |
Posted: Wed Mar 15, 2006 5:00 am Post subject: |
|
|
 Poobah
Joined: 15 May 2001 Posts: 7722
|
Sure. Because the app, and/or JMS performs slowly, apps tend to start up multiple instances of their app to handle occasional bursts of high volume. All my .NET apps get by fine with a couple instances of a SVRCONN channel running. All the JMS ones frequently have between 50-100 instances running (multiple MDBs running on multiple servers in both data centers, all client connected to the one QM).
They don't need you to define 40 separate SVRCONNs. They just need 40 instances of one SVRCONN. Of course, they might want to have a dedicated SVRCONN for each instance. Other than the initial extra 5 minutes it would take you to define these, no big deal. Benefit is then you could selectivly shut anyone down if you want. _________________ Peter Potkay
Keep Calm and MQ On |
|
Back to top |
|
 |
csmith28 |
Posted: Wed Mar 15, 2006 5:31 am Post subject: |
|
|
 Grand Master
Joined: 15 Jul 2003 Posts: 1196 Location: Arizona
|
Thanks Peter.
I knew I didn't need different channel definitions. _________________ Yes, I am an agent of Satan but my duties are largely ceremonial. |
|
Back to top |
|
 |
fjb_saper |
Posted: Wed Mar 15, 2006 5:46 am Post subject: |
|
|
 Grand High Poobah
Joined: 18 Nov 2003 Posts: 20756 Location: LI,NY
|
csmith28 wrote: |
Thanks Peter.
I knew I didn't need different channel definitions. |
If you need some more "number" crunching on JMS connection may I suggest you read the "JMS Topologies" pdf (search on IBM). It covers exactly the point you were asking, namely how many handles for what kind of operations.
Enjoy  _________________ MQ & Broker admin |
|
Back to top |
|
 |
jefflowrey |
Posted: Wed Mar 15, 2006 5:55 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002 Posts: 19981
|
Or why "MaxChannels=100" is no longer a more than adequate default in a lot of cases.... _________________ I am *not* the model of the modern major general. |
|
Back to top |
|
 |
csmith28 |
Posted: Thu Mar 16, 2006 3:54 am Post subject: |
|
|
 Grand Master
Joined: 15 Jul 2003 Posts: 1196 Location: Arizona
|
Thanks guys. _________________ Yes, I am an agent of Satan but my duties are largely ceremonial. |
|
Back to top |
|
 |
|