[
https://issues.apache.org/jira/browse/ARTEMIS-681?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15423035#comment-15423035
]
clebert suconic commented on ARTEMIS-681:
-----------------------------------------
But this is because you are load balancing every message. you should keep
messages on their origin of node. there's no way to redistribute every single
message and achieve performance.
> Do not use just one "sf.my-cluster..." queue for message load-balancing and
> redistribution
> -------------------------------------------------------------------------------------------
>
> Key: ARTEMIS-681
> URL: https://issues.apache.org/jira/browse/ARTEMIS-681
> Project: ActiveMQ Artemis
> Issue Type: Bug
> Components: Broker
> Affects Versions: 1.3.0
> Reporter: Miroslav Novak
> Priority: Critical
>
> For load balancing of messages in cluster is used just one "sf.my-cluster..."
> queue which is used for routing messages to remote cluster node. This is
> major performance bottleneck and scaling up blocker in case if there are 100s
> queues/topics on server and there is just one queue to route messages to
> another node in cluster.
> Expected behavior: There should be special "routing" queue for every queue
> and topic subscription per "remote" cluster node.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
