[rabbitmq-discuss] RabbitMQ sharding supprot

Yair Ogen yairogen at gmail.com
Mon Jun 10 06:59:22 BST 2013


I've tested out the clustering and HA support (including replication 
policies) and was successful in achieving the following:

3 nodes in a cluster with full replication on all exchanges and queues. 
Publisher sends messages to node A. I see these messages in replicated 
queues on node A, B and C. That's good.

I started 2 consumers one reading from Node B and one from Node C. They 
read the messages in a  round-robin like fashion (i.e. once one consumer 
read the message the replicated message in other node was gone).

That's all very good. 

Problem is that this doesn't scale very well. Assume I have a very high 
load from publishers. Consumers can't read fast enough. Eventually - I'll 
run out of physical resources.

Does rabbitmq support any kind of sharding or message distribution based on 
a known key?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20130609/7b86c2e5/attachment.htm>

More information about the rabbitmq-discuss mailing list