a small note on redpanda. Is not a `rewrite`. It is an entire new storage engine, different consensus protocol, different on-disk format, different internal RPC, no external dependencies, ...., in fact when i started it it was using flatbuffers as the API. What we did later to make money is add a kafka-API translation layer, but in terms of code size, is a very small part. The way i'd recommend thinking about it is a new storage engine that is compatible w/ kafka clients. https://github.com/redpanda-data/redpanda/tree/dev/src/v/kaf... - this is basically all the kafka-related code, the rest is net-new design.