If you want an asymmetric persistence layer where all writes are documents placed in a Kafka topic, and all reads are via relational queries.

It's a distributed, eventually consistent, highly available ledger that can be queried with SQL.