[rabbitmq-discuss] Distributed RabbitMq

[Michael Klishin]

On 21 Mar 2014, at 08:10, ratheesh kannoth <ratheesh.ksz at gmail.com> wrote:

> 1)  In distributed Rabbit, is there any need to run  Broker ?   if
> yes ,  where we need to run it ?
> 
>                if no,  who is responsible for queuing and storing of
> messages ?

If you use clustering, a cluster is formed by 2 or more RabbitMQ nodes.
Federation and shovel plugins connect 2 or more RabbitMQ nodes or clusters.

> 2)  In distributed  RabbitMq, we can send  messages from subscriber to
> consumer directly ?

Messages are published to and consumed from RabbitMQ nodes. If your publisher
and consumer are connected to the same node, you get the best inter-cluster
data locality but otherwise, you can publish to and consume from any node
and RabbitMQ will pass messages to the “right” node under the hood.

See http://www.rabbitmq.com/distributed.html

MK

Software Engineer, Pivotal/RabbitMQ