If appropriate, the result set of proper select requests will be cached in the   GW. If the GW detects that the same select request, together with its   

parameters has been reissued, the GW will return the result ret immediately

instead of forward it to the read-only cluster node. In this way, it can lift

the database performance greatly, with improvement from at least ten times to even serveral thousands times quicker than before.

   All is automatic, no need for source code modification.

      DBTwin Active-Active Database Cluster Diagram