[
https://issues.apache.org/activemq/browse/SMX4NMR-25?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=46165#action_46165
]
Guillaume Nodet commented on SMX4NMR-25:
----------------------------------------
The best way to do clusterting and HA is to define explicit jms JBI endpoints.
As you said, it provides much better performance, much more control over the
QoS. If you invert this sentence, you have the drawbacks of the jms flow ;-)
Another thing is that people tend to think that because it uses JMS, the JMS
flow might provide some guarantee that no messages will be lost, which is wrong
because there is no transactions involved (I would agree the risk is reduced
when using persistence though).
Usually, when you need some clustering or transactional behavior, you only need
a single pair of JMS endpoints at the beginning of you JBI application when
requests comes in and you can therefore ensure both a transactional and
clustering behavior.
> JMS flow
> --------
>
> Key: SMX4NMR-25
> URL: https://issues.apache.org/activemq/browse/SMX4NMR-25
> Project: ServiceMix NMR
> Issue Type: New Feature
> Reporter: Guillaume Nodet
> Assignee: Rob Davies
>
> Implement a JMS / JCA flow. Note that the JMS flow in 3.x is not really
> reliable and does not support redelivery / rollback / transactions.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
