On 03/17/2011 10:32 AM, Ian Barber wrote:
* JMS, and what length of stick should be used when dealing with it.
This one is rather interesting.I believe that 0MQ can't be wrapped to provide JMS API, but JMS can be wrapped to provide 0MQ-like functionality (i.e. 0MQ messaging patterns).
While that's bad news for those who would like to move existing JMS apps to 0MQ without modifying the code, it's good news for those who would like to see JMS systems (WebSphereMQ, Tibco EMS, ActiveMQ, SonicMQ, HornetQ etc.) interconnected with 0MQ systems.
Specifically, if existing JMS applications use well-defined messaging patterns (REQ/REP, PUB/SUB) -- which many of them do -- they can be incorporated into 0MQ topologies; the JMS broker basically becomes a 0MQ-style device.
Technically, JMS API has to be wrapped in such a way as to provide a 0MQ Java API. When done simple Java device can be used to bridge JMS client library with 0MQ library (see the attached diagram).
Martin
<<attachment: jms.png>>
_______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
