Question about implementation of Broker Clustering in SJSMQ

I was reading Chapter 4 of the Technical Overview of MQ 3.x:

http://docs.sun.com/source/819-2574/broker_clusters.html

However, I am still not sure if I understand some things correctly. So here are my questions:

*) Assume there are 3 brokers: X,Y,Z

*) Assume there are no consumers connected yet.

*) Then assume that a producer A connected to its home broker ( say broker X ) and produced a message M on queue Q.

1) Is the message M on queue Q "replicated" across all available nodes in the cluster ?

2) If answer to [1] is yes .... is the replication "synchronous" ? That is, broker X will wait for confirmation from broker Y and Z that they have a replica of the message from broker X ?

3) If answer to [1] is no.... and broker X goes down, then consumers started connecting to the remaining brokers ( Y and Z ), I presume this means that the message M produced on queue Q by producer A will actually not be consumed ?

4) If answer to [2] is yes, and broker X is started again, does this mean that only now will broker X tell broker Y and Z about the message M on queue Q, because broker Y and Z now have consumers ( whereas they did not have one before ) ?

Other questions:

5) Master broker: Am I reading it correct that a master broker is NOT required if I only have auto-created destinations ?

6) Is broker clustering available in Open Message Queue 4.0 ? The home page says so:

https://mq.dev.java.net/

but I could not figure out the differences with regards to broker clustering, so I am in the middle of whether to use Open Message Queue 4.0 or SJMS 3.6

7) Does broker clustering require JDBC persistence ?

Thanks

[1725 byte] By [j.salvo] at [2007-11-14]