What is the generic structure of a WebSphere MQ cell?

0
votes

What is the generic structure of a WebSphere MQ cell?
To give you an idea of what I am looking for, let me give an example of an WebSphere Application Server topology.
following are the rules for WAS

A Cell in WAS

  1. will contain one Cell Deployment Manager Node.
  2. will contain one Deployment Manager
  3. will contain at least one node.
    1. each node will contain one node agent
    2. each node will contain at least one server
  4. may have one or more clusters
    1. a cluster will have at least one node as a member

The following is a image showing the above. WAS cell structure

I have borrowed the image from an explanation given by rafaelri websphere concepts: cell, node, cluster, server…

What is really want are the rules for a similar topology structure for WebSphere MQ.
I know that there is a concept of a Cell in MQ, similarly there is a concept of a Cluster as well. I just don't know what the hierarchy of the structure is.
Any help will be much appreciated.

1
websphereibm-mqmqtopology
When you say that you know there is a concept of a cell in MQ, do you mean SCOPE(CELL)? Other than that I can think of no cell concept in MQ. MQ also has clusters, several meanings of the word in fact. Will write an answer for that part of your question. – Morag Hughson

1 Answers

0
votes

IBM MQ has the concept of 'clusters'. The term is unfortunately rather overloaded!

You can use MQ in an HA Cluster. This is the closest equivalent to your use of the word 'cluster' in a WAS environment. An HA Cluster is used for High Availability and failover of the queue manager.

Most people when they talk about 'MQ clusters' however, are referring to MQ Queue Manager clusters which is something rather different. MQ Queue Manager clusters are a set of co-operating but otherwise completely distinct shared-nothing machines which provide a highly available set of services (by advertising queues of the same name) to process your messages. Queue Manager clusters are used for Hig Availability of your service by providing multiple instances of it (the queue) and workload balancing the messages to available instances of the queue. It is not for High Availability and fail over of the queue manager.

The combination of a cluster of queue managers, each hosting a queue, which is the endpoint for your service, where each queue manager is also covered for fail over scenarios by using some kind of HA Cluster (either the built in service from MQ known as Multi-instance QMGR or an HA Cluster managed by an external HA manager, e.g. Veritas Cluster or HACMP) is best practice for overall availability of your messaging system.

Here's a diagram that shows the structure of what I've described above. You'll note that there is no cell concept.

IBM MQ structure