Internally the very bottom of the stack on the hosted product is actually just a write ahead log on RDS and in the open source product it's just sqlite. We'll likely eventually build our own durable write ahead log, and have plenty of experience doing so, but this hasn't been needed so far.

It doesn’t use Kafka, the transaction logic is custom implemented in Rust.

Convex does not have special support for unique constraints, as they are easy to implement in your own code (like you wrote, the mutations are happening “inside” the database). This follows the philosophy of making it clear what the database primitives are, and hence what performance you can expect from the database.

Of course since your code is all JavaScript/TypeScript, it’s easy to layer in constraints via a library, like in https://labs.convex.dev/convex-ents/schema#unique-fields