[rabbitmq-discuss] Can I set the level of replication for rabbitmq clustering?

[Alexis Richardson]

Gautam

Have you considered managing the work load by migrating consumers
instead of queues?  This is not only substantially less expensive, but
it's also much better supported by Rabbit.

To your question: while it is possible to manage the number of
replicas, this is for mirroring and not migration.  Therefore, even if
you assume zero network cost, you will still need multiple consumers
on what is in effect a single logical queue.

alexis


On Mon, Apr 30, 2012 at 9:54 PM, Gautam Bakshi <gautam.bakshi at gmail.com> wrote:
> Hi Everyone!
>
> My work items are very small but amount of work is very large  and network
> IO is quite a large overhead(not Rabbit's fault, my bandwidth is limited).
> So I'm thinking of running a local rabbitmq on each of my worker servers and
> then replicating work locally for them to pull and push messages to.
>
> I can write a program that messages the queue by monitoring them and if one
> servers queue gets empty it takes a few thousand from another server but I'm
> wondering if thats needed if I use the clustering feature? I know how to set
> it up but how it works is a bit unclear to me.  Can I set the replication
> not to replicate everything all the time but only a few jobs when the queues
> get below a specific threshold?
>
> Also, right now I'm using localhost to connect but if all my access is
> entirely local is there a faster way to connect such a direct socket?  Any
> suggestions or ideas would be greatly appreciated.
>
> _______________________________________________
> rabbitmq-discuss mailing list
> rabbitmq-discuss at lists.rabbitmq.com
> https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
>