MQSeries.net :: View topic

kbsinc · Posted: Thu Feb 06, 2003 8:39 am Post subject:

I am in the beginning phases of architecting a solution involving clustered QMgr's interfacing with non-clustered QMgr's. We wish to take advantage of the workload balancing and failover aspects of the clustered approach for our application. After reviewing many of the messages posted in this forum and after consulting the WebMQ QMgr Clusters manual I am still left with a fundamental questions that I am researching.

Proposed environment: QMA1 QMA2 QMA3 QMGW are V5.3 QMgr's in a cluster. All running on separate AIX boxes. QMA1 and QMGW hold full repositories, QMA2 and QMA3 a partial repository. QMA1 QMA2 QMA3 all have an identical copy of the application installed. QMGW would be considered the 'gateway' into the cluster for applications connected to QMgr's that reside outside the cluster.

Q.) I would like a message that is sent from outside the cluster to be able to be processed by any one of the application QMgr's (QMA1, QMA2 or QMA3) using QMGW as the connection point into the cluster and the round robin approach of the workload exit in place on QMGW. Am I correct in my assumption that this is the best (and possibly only) approach to get applications connected to non-clustered QMgr's into the cluster so that their messages can be processed by any one of the clustered application QMgr's?

Q.) Is there any way to lessen the impact of QMGW being a single point of failure for the non-clustered QMgr's?

As I can continue my research into this topic I would be interested in any insight from some of you who may have encountered a similar architectural issues.