An application client can run on a system anywhere in the WebSphere® MQ network. If your applications use WebSphere MQ facilities to connect to the broker, and to interact with it (using the MQI and AMI), you must set up the WebSphere MQ resources that they require.
The way that you set up applications is identical to that for clients for an WebSphere MQ server. To support client connections to a broker:
An application can get messages only from queues that are owned by the queue manager to which it is connected (this restriction is true for all WebSphere MQ applications). Therefore, if an application expects to receive messages from a queue populated by a service within a particular broker and owned by that broker's queue manager, it must connect to that broker's queue manager (using a local or WebSphere MQ client connection).
An application that puts messages, however, can be connected to any queue manager in the network, provided that the queue manager can resolve the target destination in some way. In all cases, the queue manager to which the client application is connected must know the location of the queue or queues to which the application puts messages (for example using remote queue definitions).
When you define a WebSphere MQ queue as a node for a message flow, you must not give it a name that starts with SYSTEM_BROKER. Names that include these characters are reserved for queues that are defined for internal use by WebSphere Message Broker components.
If your application is a subscriber that receives messages published by other applications, it can specify a temporary dynamic queue as its subscriber queue. If it does so, the broker automatically deregisters the subscription when the queue is deleted.
For more details about applications, putting and getting messages, and the use of WebSphere MQ clients, see the Clients and Application Programming Guide sections of the WebSphere MQ Version 6 information center online or the WebSphere MQ Version 5.3 book on the WebSphere MQ library Web page.