[rabbitmq-discuss] Alice in Wonderland chasing the RabbitMq - Email found in subject
jvarona at directpartners.com
Sat Jul 16 00:14:06 BST 2011
So things may have changed with the implementation but the last time I worked with Rabbit clustered nodes simply brokered requests to the master node. So depending on which node you connected, that connection brokered a request to the node where the queue existed. The notion of replication with full transactionality did not exist since 1) there weren't replicas of data on at least 2 nodes and 2) since there were not replicas, transactionality did not have to span multiple boxes. If this has been corrected, and please forgive me since my testing and review of all the code was circa 2009. It probably means that the throughput of Rabbit has probably diminished a great deal. Is this not the case anymore?
From: Gavin M. Roy [mailto:gmr at myyearbook.com]
Sent: Friday, July 15, 2011 1:54 PM
To: Jorge Varona
Cc: rabbitmq-discuss at lists.rabbitmq.com
Subject: Re: [rabbitmq-discuss] Alice in Wonderland chasing the RabbitMq - Email found in subject
Serious problems with replication? What are you replicating? Other than HA, I'm having a hard time envisioning a desire to replicate queues since there are multiple routing models to publish data across multiple nodes without extra effort on your part.
I've scaled Rabbit to a fairly high velocity without any scaling issues. Not sure what the gripe is here.
On Fri, Jul 15, 2011 at 3:43 PM, Jorge Varona <jvarona at directpartners.com<mailto:jvarona at directpartners.com>> wrote:
I have to ask, why go with RabbitMQ. There were some serious problems with replication last time I checked, which could make it hard to scale.
rabbitmq-discuss mailing list
rabbitmq-discuss at lists.rabbitmq.com<mailto:rabbitmq-discuss at lists.rabbitmq.com>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the rabbitmq-discuss