Clustering provides the mechanism to group more than one server, Queue manager as a whole so that client treats as a single entity.

Advantages :-

• Failover
• Load-balancing
• Scalability
• Easy to administrator

How to achieve clustering in Websphere queue manager

There is a concept of repository which cluster software maintain about the meta-data of servers, queue managers, shared queues, etc.

Repository is nothing but type-specification of queue-manager.

Types:-

• Full repository
• Partial repository

Full Repository :- holds cluster configurable information. If queue manager marked at full repository in cluster went down then whole cluster went down. So, at the design time better to assign 2 queue manager as full repository so that another will take charge.

Partial Repository :- QM marked as partial repository acts as client of full repository. If this need any meta-data or configuration , then it ask to full repository QM though command queue.

Note :- more than 1 queue manager marked as queue manager do syncing among each other, so if we make more than 2 QM as full repository then lots of syncing take computer processing resulting performance degradation. So, best is 2 QM assigned as full repository. Other will be partial.

Steps involved in clustering.

• In your enterprise, identify the at most 2 queue manager which will be marked as full repository.
• Define cluster-sender and cluster-receiver channel in full repository.
• Define a cluster local queue.

Create a Queue manager Test1 (TCP Port 1420)

Create another queue manager Test2 (TCP port 1421)

Create a cluster sender channel named TO.TEST2 in TEST1 Queue manager

Create a cluster receiver named TO.TEST1 inside TEST1 queue manager

 

Create a cluster sender channel named TO.TEST1 inside Test2 queue manager

Create a cluster receiver channel named TO.TEST2 inside Test2 queue manager

Create a cluster named TESTCLUSTER

Create another queue manager Test3 and assigned it as partial repository