OpenJPA version of Correlator implementation is not scalable
------------------------------------------------------------
Key: ODE-428
URL: https://issues.apache.org/jira/browse/ODE-428
Project: ODE
Issue Type: Improvement
Components: BPEL Runtime
Reporter: Sean Ahn
Whereas the hibernate version uses a separate table, bpel_unmatched and does
exact-match sql queries using the correlation key value and correlator id, the
openjpa version does not have a corresponding table and loads up all message
exchanges for a correlator and compares the correlation key values in memory.
The openJPA version does the following details:
1. A correlation key for an incoming message that does not have a receive
operation at the moment is saved into the message_exchange table.
2. Later when a receive operation is invoked, all message exchanges that belong
to the corresponding correlator are loaded up to memory from the database.
3. Then, the correlation keys in the individual message exchanges are compared
against the incoming message.
Loading up message exchanges to memory based on a correlator is a potential
performance problem since the more process instances you have, the more message
exchanges will be in the database.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
