Hyperdrive

Hyperdrive makes it practical to use a traditional Postgres or MySQL database from Workers. It pools connections close to the database and can cache queries, so each invocation does not pay to set up a connection across an ocean.

Setup takes minutes: give it a connection string and swap in the binding. Read-heavy workloads can see an immediate latency improvement. Including it with paid Workers plans removed the last reason to hesitate.

Points off only for opacity: you get limited visibility into pool behavior, cache hit rates, and where your pool is actually placed, and the config surface is minimal when you want to tune. But as the bridge between “we have a real database” and “we want to build on Workers,” it’s exactly the product it should be.