I am quite new to JBoss clustering, and have problems understanding these issues:
I read that mod_jk2 connector has fail-over and session replication capabilities. Why would I need to eneble JBoss clustering if jk2 will distribute the load over the dif. app. servers, and makes sure fail-over happens (and replicates the session) when one of them fails??
(This assumes I have tomcat-Jboss bundled, and not clustering at both Web container (tomcat) and EJB container (jboss) )
Assuming EJB home object are cached, to improve performance, and not rely on lookup for each call(Service locator/EJBhome factory pattern). What happens if the Cached bean references to a Server that goes down. (is there a Fail-over mechanism at the client stub, like is the case for WebLogic?)
And by the way, I don't want to have the front end Apache HTTPD to be the sinle point of failure, but don't have Money/time to invest in a load balancer like Linux Virtual Server/Cisco L4-L7 Switch. Is there a module (just like mod_jk that connects to Tomcat) that connects 2 apache instances together, and provides failover when one of them is dead?
Thanks A LOT.
Any help will be very much appreciated! (please give your advice even if you don't have all the answers!)