I'm working on a use case where a single workflow in IBM App Connect needs to send requests to multiple services in parallel. The requirements are:

• The workflow should not wait for one service to respond before calling the others.

• Each service must eventually succeed - if a request fails, it should be retried automatically until it completes successfully.

• The entire workflow should only be marked complete when all services have returned success.

I'm looking for guidance on the best design pattern or App Connect features to use for this scenario.

• Should I be using parallel flows or callable flows for this?

• What's the recommended way to implement retries with backoff in App Connect?

• Is there a way to track overall completion across multiple parallel service calls?

Any advice, examples, or best practices would be greatly appreciated.

Thanks in advance!

Hello @Ali Mashally

Callable flows are not an alternative to parallel flows, they are a type of flow designed to provide the ability to split flow logic in hybrid deployment patterns (I could invoke multiple callable flows in parallel for example).

With regards to parallel invocations and co-ordination across the requests, have you looked at the aggregate and group nodes (https://www.ibm.com/docs/en/app-connect/13.0.x?topic=nodes-group-aggregation)? From your description here, these nodes and associated patterns (https://www.ibm.com/docs/en/app-connect/13.0.x?topic=events-using-aggregation) would appear to meet your requirements.