i am absolutely new in this area and we are going to set-up hornetq as a MOM in our production servers soon. after reading the documents, i was thinking that it will be good to have client side fail-over with 2 or more hornetq servers in clustered/ non-clustered mode (load balancing may not be a big concern for us). we will be using persistent data store. but using SAN is not an option for us, and using NFS as a shared storage for live/ back-up servers probably turns out to be a synchronous system which invalidates one of the most important feature of asynchronous messaging. for data safety, i think using RAID-5/10 in each of the server seems to be sufficient. So when a system/ disk crashes, durable data should be recovered in some time (after replacing the disk) and till that time, the other server can respond to both the clients. wanted to know if i am missing any crucial points which might affect the flow severely.
Your question is not clear enough to get an accurate answer, but I can say the following: