[rabbitmq-discuss] RabbitMQ HA master/slave performance

[Michael Klishin]

2014-03-09 23:49 GMT+04:00 mlabarinas <marianogabriel at hotmail.com>:

> But this is something that I don't control. Would I have to set my
> producers
> to produce to a node (master maybe) and my consumers to consume in others
> (slaves maybe)?
>

Best data locality is when you publish and consume from the master.

Some set up load balancers and make their clients connect to it.
This makes queue masters distribution more even and, in general, improves
data locality.

There are also projects that use HTTP API to discover queue masters and make
consumers connect to their respective nodes.
-- 
MK

http://github.com/michaelklishin
http://twitter.com/michaelklishin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20140310/eab9fd13/attachment.html>