Dear JBoss-Community ,
i have some issues with the MessageStore Strategy of the ESB, which i really want to discuss with you.
1. First of all why must their be just one global MessageStore. I think for many application this strategy doensn't fit.
Of course an audit trail can be done easier using one message strategy but in my opinion fail over strategies are more complex to handle for specialized services.
2. The routers or better the ServiceInvoker (implied through issue 1) only accepts the global DQL or RDLVRQ Strategy which are globally configured
within jbossesb.esb. For async routers it is completely impossible to integrate a Custom MessageStore, because they are using all the predefined DLQ and RDLVRQ Services (jbossesb-esb).
My suggestion is to make this behaviour more generic, because I think in real application it is impossible to have just one big MessageStore for failed messages.
Furthermore i think there must be made a differnece between Audit Trail reqirenments and fail over requirements.
Within my current project i implemented a new Redeliver MessageStore and a new Redeliver Scheduler which is configurable. unfortunately the Redeliver Servicea must only use
ServiceInvoker.deliverSync method which is not desirable for esb projects.
I hope for arguments which can explain the decision to be completely dependentend of a global MessageStore and especially the Router (StaticRouter) behaviour.
Thank you very much.
Best Regards yours sincerely
Christian
