Hello Andriy,

Thanks for the questions and apologies for the delay in responding.

(1) Yes, even if you are spinning a brand new greenfield project, using Debezium will offload many rudimentary things from you such as exactly once delivery of events to Kafka, governance over event schema evolution, serde, snapshotting, etc.

I think it is always worth leaving them to a middleman while focusing your attention to business logic.

(2) Nothing specific about using Materialize here. I just wanted to build and example integration to showcase how these technologies can work together.

If you are from a JVM background, using Kafka Streams would be your first choice here.

(3) Pinot guarantees the data freshness upto a couple of seconds. Also, we are not doing a "super latency-sensitive" use case like anomaly detection here.

(4) Yeah, that name could've been avoided. I will make sure to do that from the next time.

Thanks for you comments and hope I answered all your questions reasonably well :)

Regards,

Dunith

(2)