Nice blog. It is a clear trend that more real time analysis applications are being built, and you don’t want to write the program twice, once for streaming and once for historical data backfill. Won’t be nice if there is a single data platform allow you to query either batch mode or streaming mode, or hybrid mode(in a subquery to query historical data while the outer query is to continuously processing incoming data) Paimon is more for JVM ecosystem. We at Timeplus open-soured a pure C++ “streaming OLAP” by extending ClickHouse with a Flink-like query engine and a Kafka-like streaming storage. https://github.com/timeplus-io/proton Maybe you can check out what the heck is Proton 😎