[rabbitmq-discuss] RabbitMQ server to server connection

Tue Aug 5 09:44:25 BST 2008

Manuel,

On Tue, Aug 5, 2008 at 8:01 AM, Manuel Duran Aguete
<manuel+rabbitmq at aguete.org> wrote:
> Enterprise A has a mq installed, Enterprise B has another mq installed, an
> application in A want to request some information and send a message to a
> queue in B. The mq in A routes the message to B usign some criteria.
> Application in B responds to A, by sending the response throught B.

One option to do the same things as transmission queues is described
here: http://hopper.squarespace.com/blog/2008/6/22/introducing-shovel-an-amqp-relay.html

Having said that, you may want to ask yourself whether you need to do
things in the same as MQ series.

An alternative to this, which may or may not suit is B to be a plain
consumer on the broker inside A.

This would be a simple asynchronous RPC pattern.

Although I'm not pretending to be an expert on MQ series, some of the
reason why different enterprises use transmission queues is because
the MQ series client doesn't work over a WAN very well, so people tend
to notice a lot more reliability and efficiency letting a two QMs
forward messages to each each. I think encryption, authentication and
demarcation of responsibility play a role as well.

AMQP, on the other hand, has been designed partly with remote
connections in mind so as a plain jane AMQP client, you can consume
directly from a "remote" queue and publish directly to a "remote"
exchange.

On the other hand, one advantage of relay queues (or transmission
queues in MQ parlance) is that do allow you to buffer and compress
messages going across a WAN, which may be an efficiency gain.

HTH,

Ben