> each Kafka partition was represented by a separate SQLite database
What does the programming interface look like for this? I believe you'd only be able to use SQL to query each partition (database) individually, unless there's another layer orchestrating SQL over the partitions. I wonder if you just end up with connectors again.
