CallRouter issues

Question

          Are there any dicussions or guidelines on the use of CallRouter based routing?
          I am exploring a design where the home interface of an entity bean is load balanced
          and the entity bean object is not load balanced but can fail-over. I.e. the collection
          of entity beans are instantiated on multiple servers, but any one entity bean
          (identified by its primary key) is not instantiated on multiple servers simultaneously.
          I am thinking of implementing a call router algorithm that returns 2 servers,
          effectively a primary and a backup. For any one entity object, calls should be
          routed to the primary and the secondary used only if the primary fails.
          Is the order of server names returned from getServerList significant and is it
          guarenteed that the second server in the list will only be used if the first fails?
          Is there any more detailed documentaion on the Cluster API?
          

Answer 1

          I think you are implementing the behavior of a stateful or entity
          session bean that is not read-only. You may wish to investigate
          that before spending too much time with Call Router.
          Mike
          >
          >I am exploring a design where the home interface of an entity bean is
          >load balanced
          >and the entity bean object is not load balanced but can fail-over. I.e.
          >the collection
          >of entity beans are instantiated on multiple servers, but any one entity
          >bean
          >(identified by its primary key) is not instantiated on multiple servers
          >simultaneously.
          >
          >I am thinking of implementing a call router algorithm that returns 2
          >servers,
          >effectively a primary and a backup. For any one entity object, calls
          >should be
          >routed to the primary and the secondary used only if the primary fails.
          >
          >Is the order of server names returned from getServerList significant
          >and is it
          >guarenteed that the second server in the list will only be used if the
          >first fails?
          >
          >Is there any more detailed documentaion on the Cluster API?