Hi. I would like to join two stream and perform window aggregation. One stream will be transaction data and another one is fact stream. Is there a way to trigger window aggregation based on time and some event from my fact stream. For example, if i update the area of my service, i need to recalculate the whole window for that specific area.

With Flink SQL, windows can only be applied to append-only streams. There’s no support for windows that might need to recalculate/update their results based on updates to the stream(s) fed into the windows. That sort of windowing needs a different (and much more expensive) approach to managing window state. You could probably implement this using the DataStream windowing API, but it could get very complex, depending on the exact requirements. You could definitely build this using process functions.

Okay thank you so much.