In most readying around kafka streams it is not recommended to you request/response within a kafka stream application, or more generally interacting with external system synchronously within a stream topology.
However i have a case where i need consume a stream of claims that reference documents stored in S3 (one for each claim) and process it to produce small record. Essentially a candidate for a FlatMap.
I wonder if anyone have experience with this kind of scenario using kafka streams and what has been the outcome.
If discouraged, what would be an alternative solution ?