On 9/4/07, jgunz <[EMAIL PROTECTED]> wrote: > > I've been reading through the Camel documentation and about its incorporation > in ActiveMQ 5.x at > http://activemq.apache.org/enterprise-integration-patterns.html. I think the > combination is going to be very useful to me and am just brainstorming about > how I am going to apply both ActiveMQ and Camel in an upcoming project.
Cool! :) > How will the EIPs work with clustering? In particular, I'm wondering if > there will be any issue if I install a simple routing rule to copy messages > from topic foo to topic bar if I have a cluster of several ActiveMQ brokers > (each of which will each probably have the routing rule installed). > > I'm not all that familiar with how clustering works so this may be a dumb > question. The EIP patterns would run above the clustering layer. For example if you have brokers in a master/slave arrangement to give you HA; so that messages are replicated in multiple physical locations in case of serious hardware failure - from the Camel perspective, its just normal message flows across JMS (using declarative Spring transactions for ACID). e.g. a routing rule that copies message from 'foo' to 'bar' would logically be deployed in both master and slave brokers; but the slave rule would only become active when the slave becomes the master. The only thing to be careful of is that each broker in the cluster (or network of brokers if you are using store and forward) is running the same routing rulebase. Right now the routing rules are defined either in a jar or a spring XML file; in the future we could come up with some kind of cluster-wide rule base for folks who want highly dynamic routing rules to be applied across the cluster in a uniform way. Until then we just need to make sure that each broker is running with the correct routing rules jar or Spring XML. Maybe if we turned ActiveMQ into an OSGi bundle it might be a bit easier to manage running a number of brokers in a master/slave cluster with exactly the same classpaths / Camel XML & reduce the possibility of inconsistencies. -- James ------- http://macstrac.blogspot.com/
