I am working with a group that has an application that has a very critical need for high availability and performance. To make a long story short, they want to use DB2 HADR for availability. But they also want to use the DB2 HADR read-on-standby facility if the application server is co-located with the database server, for performance reasons.
The above details aren't really important here. But what I would like to know is if there is a way to dynamically choose which datasource the JPA calls will be using at runtime, based on some algorithm. All update calls to a JPA entity will always use the primary datasource. But if the entity is only being read, then the standby datasource may be the one used if the servers are co-located. I'm not sure if this is even feasible with JPA, but hopefully someone here can tell me. Thanks!
