0 Replies Latest reply on Aug 11, 2009 12:05 PM by Tai

    clustering using JBoss Pojo Cache and JTA without using an A

    Tai Newbie

      Hi,

      we are evaluating JBoss Pojo Cache and JTA/HA-JNDI for clustering and global transactions to use within our server (agent) implementations. Our agents holds a data container. The goal here is to cluster our agents on different nodes and JVMs in a cluster by having an agent's data container replicated in each cluster's node.

      The requirements for the cluster are:
      - fail over solution: currently client requests uses RMI to access the agent. In case of a graceful/disgraceful leave of an agent node the client requests should be handled by another node within the cluster
      - (at the beginning simple) load balancing: read operations should be handled by one agent node and write operations by another agent node.

      Since this would be a major change for our architecture we do not want to use an application server in the first place. Instead we want to keep our agents to run stand alone and being cluster and JTA enabled.

      The questions are:
      - is there anybody experiencing with clustering using JBoss Cache and JTA without deploying it in an AS?
      - it seems that for global transactions and fail over HA-JNDI is a requirement for JTA. As far as we could see HA-JNDI is tightly coupled to JBoss AS and cannot be used stand-alone. Correct?
      - Is there a HA-JNDI standalone solution?
      - Is there another stand-alone alternative for HA-JNDI?
      - Is there a stand-alone load balancer available for doing both: load balancing and re-routing request in case a node crashes?

      Thanks in advance
      Tai