Compatible with MySQL or PostgreSQL is becoming a big deal. This is great for users but there will be confusion about the meaning of "compatible".
smalldatum.blogspot.com/2021/10/compat
Conversation
Replying to
I classify systems with pg compatibility based on code reuse. The extreme case is a pure extension - these "reuse" everything. Then you have Aurora, which can reuse existing geospatial + FTS index AM code. Cockroach/Yugabyte change storage at a higher level, making this hard.
3
5
Cockroach/Yugabyte do still reuse a lot of pg code that doesn't touch storage. Don't know where to place new Google spanner stuff just yet -- maybe it's kind of a clean room reimplementation of the parser. If so, that seems like another gradation in compatibility to me.
1
1
CockroachDB is reusing pg code? I thought it's pretty much written in Go?
1
1
Looks like you're right about CockroachDB - my mistake. I must have lumped it together with Yugabyte because they're similar in other ways.
Yes, similarity is distributed storage based on Google Spanner paper. Difference is CRDB builds a SQL API compatible with PG syntax. YBDB plugs the distributed storage to a PG fork (in table/index access method + some additions in planner/optimizer)
1