info wanted about JMS clustering

[marlow.andrew]

I am working on a communication mechanism between 2 systems based on
the use of JMS. My initial design employed a number of Qs to maximize
opportunities for parallelism on both the enqueeuing and dequeueing
side. The JMS implementation that I plan to use is ActiveMQ. I have
been told that ActiveMQ has a clustering facility that means I do not
need to code for multiple Qs, it is a matter of configuration. That
sounds great but my question is "is this facility a std part of JMS?".
I am concerned about implementing a solution that might depend (for
its scalability) on a particular implementation of JMS. I don't want
to do that if I can help it.

Regards,

Andrew Marlow

 

[Arne Vajhøj]

I am working on a communication mechanism between 2 systems based on
the use of JMS. My initial design employed a number of Qs to maximize
opportunities for parallelism on both the enqueeuing and dequeueing
side. The JMS implementation that I plan to use is ActiveMQ. I have
been told that ActiveMQ has a clustering facility that means I do not
need to code for multiple Qs, it is a matter of configuration. That
sounds great but my question is "is this facility a std part of JMS?".

No - it is not.

JMS is a pure interface.

I am concerned about implementing a solution that might depend (for
its scalability) on a particular implementation of JMS. I don't want
to do that if I can help it.

Study the JMS spec and the JMS java docs to ensure that you use
a standard API.

The rest is implementation specific.

It would be very reasonable to put in your docs that to achieve
redundancy a redundant message queue implementation must be used.

Arne