Here is an intriguing question that I am faced with. I have a trading partner that sends us EDIINT messages with order information within them. Some messages have a single order, others have a large batch.
When receiving several large batches in sequence our server grinds down to intolerable processing levels. I’m thinking to alleviate this problem we would have to stop TN from doing the splitting of the EDIINTs. However I only want to do this for the batched orders for this particular trading partner, can this be done by using a TPA or does it have to be a server wide setting? And typically how would one go about doing this? (Point me to relevant doco and I’ll still be happy).
Johnny, please check wM TN Users Guide for information on Large Document Handling.
But there is also a scenario large doc handling won’t help. It occurs when several larger docs (under the large doc threshold) arrive to be processed.
The EDI process splits all the groups and transaction sets. For each, another thread might be spawned, depending on the processing rule.
What’s needed in this case is a way to throttle the processing or spread the workload.
IMO, stopping TN from splitting won’t help. Ultimately the interchange needs to be split and whether TN does it or another service does it, the same performance hit will be encountered. But if you want to, splitting is indeed controlled by the TPA settings and is thus controlled at a partner level. Refer to the EDI for TN documentation.
As far as pursuing other approaches, I’d need to know more about how things are set up. Is the rule that processes the transaction sets sync or async? Do you have multiple IS instances that you could farm out work to?