[BadassFractal]

Take Redis for example: simple KV database with in-memory perf and a very comfortable API with option to flush data to disk periodically. Addresses a lot of interesting scenarios such as session storage, request throttling by a certain key etc. Not used as a replacement for a RDBMS most of the time, but rather as a specialized tool for a certain use case. At some point your RDBMS is already hammered hard enough, you don't need to dump everything into it, but ultimately you could, especially at first. Comes with clustering for free as long as you're aware of the risks and use it appropriately.

This is not to say you couldn't turn Postgres into something similar. Put the data on a ramfs, relax various write guarantees (lots of knobs in PG) etc.