[rabbitmq-discuss] The best practice of deploying

Simon MacMullen simon at rabbitmq.com
Tue Sep 18 17:15:44 BST 2012


In ancient times Erlang's multi-core support was not great, and we used 
to suggest installing N nodes on an N core machine.

But those days are long past, and doing the above just introduces 
additional overhead, both CPU overhead to the broker and administrative 
overhead to the sysadmin.

Just install one mode per machine.

There is a valid use case for more than one node per machine for 
learning about clustering and sometimes for testing, but not for production.

Cheers, Simon

On 18/09/12 15:26, inetfuture wrote:
> Hi,
> I am wondering which of these would be better:one node per machine or
> multiple nodes per machine(it/they serve the same application)?
> Dose one node can fully use resources of a machine? Dose run more than
> one node on a single machine will be better?
> I'm using RabbitMQ with Celery, there would be a few machines to run
> RabbitMQ nodes(in a single cluster), so I need to decide that how many
> node to deploy on per machine. I'm wondering this because in usually
> practice, like multiple processes/threads, "multiple" will do better,
> but I'm not sure how RabbitMQ dose.
>
> Thanks,
> Aaron
>
>
>
> _______________________________________________
> rabbitmq-discuss mailing list
> rabbitmq-discuss at lists.rabbitmq.com
> https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss


-- 
Simon MacMullen
RabbitMQ, VMware


More information about the rabbitmq-discuss mailing list