Performance tuning in objects involving Integration process (BPM)
We are facing a performance problem in one of our objects having integration process. The messages are getting processed sequentially instead of parallel processing within the integration process. As a result, the messages are getting queued up in the queue waiting for processing. Thus by the time all messages get processed, the time taken would have been alarmingly high. There is no correlation involved in the int process. My questions are
1. Is there any performance related best practices/checklist/config steps available for BPM related implementation. If so, please furnish/provide the links if any.
2. If it is not related to BPM , but related to tuning up of the queues, please provide the guidelines for queue tune up.
3. How to ensure that the messages are processed in-parallel within BPM ?