This is much simpler and more powerful in ActiveMQ 4.0 - you can configure different Dispatch Policies directly.

For now in ActiveMQ 3.x we have a PrefetchPolicy configuration. This allows a certain number of messages to be dispatched to a consumer before they are acknowledged. This feature is to achieve high performance; the higher the value the less likely it is that a client has to wait for messages to arrive before processing the message and so the higher the throughput.

e.g. under heavy load of a busy system you probably want to dispatch 1000 or so messages to a consumer, so as soon as its ready to receive a message, there are a bunch of them in memory waiting to be used - rather than having a slow request-response RPC each time.

However the effect of this is that when you send a small number of messages, they tend to all go to one consumer unless you've lots of messages. Note that this issue can be avoided in 4.x as there is now a configurable DispatchPolicy so that you can enforce, say, round robin dispatch policy to ensure that messages are dispatched fairly to avoid this issue.

In ActiveMQ 3.x or later a work around for this issue is to reconfigure the PrefetchPolicy on the ConnectionFactory/Connection so that the broker only prefetches a single message per consumer; though this will greatly reduce the performance and throughput of the system.

Configuring the dispatch policy in ActiveMQ 4.x

In AMQ 4.x, the dispatch policy can now be configured on the xml configuration file (ie. activemq.xml)
ie.

<destinationPolicy>
   <policyMap><policyEntries> 
      <policyEntry topic="COMMAND.>">
         <dispatchPolicy>
           <strictOrderDispatchPolicy />
         </dispatchPolicy>
         <subscriptionRecoveryPolicy>
            <noSubscriptionRecoveryPolicy />
         </subscriptionRecoveryPolicy>
      </policyEntry>
   </policyEntries></policyMap>
</destinationPolicy>