-1

We have 2 systems that exchange tickets base on a transactional flow, they have an order in the ticket statuses if one status does not reach one system all the flow is stuck. The problem is that we use a multi threaded, load balanced message broker between this systems and we might have cases when an update1 status can be processed faster than a create, or update2 faster than update1.

I'm look for a best practice for this kind of integration.

CyberDracula
  • 316
  • 1
  • 2
  • 6
  • 2
    Stack Overflow isn't really the place to ask general "best practices" kind of questions. Can you reformulate your question to include the specific technology you're working with? If you really want to ask in general, the question might be a better fit for [programmers.se](http://programmers.stackexchange.com/help/on-topic). – Jeroen Mostert Feb 24 '15 at 08:40
  • right now i am looking for a best practice, i'll try you link, thank you – CyberDracula Feb 24 '15 at 08:53

1 Answers1

0

This sounds like you need to implement the Scatter-Gather EIP:

http://www.eaipatterns.com/BroadcastAggregate.html

In IBM Integration Bus or WebSphere Message Broker you can set aggregation timeouts to make sure you only proceed with the post aggregation flow if you have all the components of the aggregation.

Any parts of the aggregation which don;t turn up can be timed out and processed separately.

Dave
  • 633
  • 4
  • 6