On 3/24/06, d j <[EMAIL PROTECTED]> wrote: > > Just a little further., Let's imagine a grid computing network for instance. > All gridlet has same queue, but identified by fqdn (full qualified domain > name). A publishing to an queue will publish to a queue with domain name. > The broker on the edge will route the message to the right destination. In > that sense of routing, it will really make the activeMQ a much bigger > player. Otherwise, it will be used in very limited system and stays nitch.
I agree that broker-specfic queues are useful, but they are not strictly required to be a non-nitch messaging system. Using demand based forwarding we can reproduce pretty much the same effect of certain queues being hosted on certain brokers by just being careful on what brokers we run JMS consumers on and then let the demand based forwarding mimick the same topology of broker-specific queues. -- James ------- http://radio.weblogs.com/0112098/
