Discussing different strategies to reduce locking in our @twimpact backend with @thinkberg. In-memory alone isn't fast enough anymore ;)