I'm using Debezium to perform CDC on Postgres. Debezium provides some metadata about each event including the transaction timestamp, transaction id etc.
I want to generate consistent aggregates using Kafka Streams by performing KStream-KStream joins with a Session Window. I'll be using a custom TimestampExtractor which uses the transaction id as the event time and hence all events from the same transaction will be joined. Has anyone tried something similar before or can let me know some alternative ways to implement this? Regards, Ashhar Hasan
