[rabbitmq-discuss] Cluster capability?

[Paul]

Hi All,

I've been away from RabbitMQ and from this list for about two years. I'm a bit rusty. And having lately begun to think about a distributed architecture, and knowing that RabbitMQ is *good* for this kind of thing, I realized that I need to ask some questions. 

My initial requirement is simple to state: if a client on computer A publishes to an exchange on computer A, can that message be auto-magically propagated to other brokers on other computers?

The reading I've done suggests that clustering (as opposed to federation or Shovel) might be the way to go.  But then I read something about HA queues and was uncertain as to what is required.

A use case here looks something like this:

1. Client on computer A publishes message X to local exchange on computer A
2. Client on computer B consumes message X from its local exchange on computer B

The intent here is that each client should need to know only its local broker. Clients should not need to know anything about another computer or broker.

How is this best done using RabbitMQ?

Thanks!

-Paul