Comment by jackfischer

Congrats guys! Curious how the read write splitting is reliable in practice due to replication lag. Do you need to run the underlying cluster with synchronous replication?