[rabbitmq-discuss] RabbitMQ general and cluster doubts

[Jason J. W. Williams]

The primary benefit to clustering (without using mirrored queues) is
scalability and publishing availability. If you put the Rabbit nodes behind
a load balancer like HAProxy, you'll get fairly even queue distribution
across the nodes. Queues are created on the node their creator was attached
to at the time of creation, and there is no re-balancing. Clustering also
ensures your producers always have a place to publish.

Mirrored queues give you the ability sustain node failure and not lose
access to the queues on that node. But you get a performance hit for the
additional overhead.

-J

On Wed, Nov 16, 2011 at 9:13 AM, Guillem Anguera <guillem at trovit.com> wrote:

> Dear RabbitMQ members,
>
> We're testing a two nodes cluster running RabbitMQ (2.6.1) and we don't
> understand the cluster benefits beyond having a shared knowledge about the
> metadata, as queues are sticky to the node upon which they are created,
> queues are limited to node's hardware capabilities (CPU, Memory, Disk,
> ...). Can you explain us the cluster benefits? After adding another node to
> the cluster, is there any way to rebalance the existing queues among all
> nodes?
>
> We're planning to set up about 30.000 queues and 100.000.000 messages with
> an average of 1 Kb each, every time a message is consumed it reenters the
> corresponding queue until it's deleted. What do you think?
>
> Looking forward to hearing from you.
>
>
> --
> Guillem Anguera
> guillem at trovit.com
> Tel: 686.528.792
>
>
>  <http://www.trovit.com>www.trovit.com
> Your partner for qualified users
>
> _______________________________________________
> rabbitmq-discuss mailing list
> rabbitmq-discuss at lists.rabbitmq.com
> https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20111116/b9832dd1/attachment.htm>