PlanetScale for Postgres
Managed Postgres where the disk sits physically inside the machine running your database, not at the far end of a network. The vendor calls it Metal: NVMe attached to the server, included in the cluster price. The rest looks like the competition, database branches, replicas, backups, and a plain Postgres client in front.
Strengths
- NVMe storage is attached to the server and included in the cluster price, so disk performance is not a separate line that keeps growing on the invoice.
- Plain Postgres in front: your ORM, your migrations and your backup tooling do not change, the exit door stays open.
- The vendor spent a decade operating databases at scale on MySQL before porting the same machinery to Postgres. This is not a team discovering the job.
Limitations
- No free tier. For a mockup, a course or a weekend, you pay where three competitors let you start at zero.
- Local disk is an operations advantage, not a product feature. Nobody on your product team will see the difference, and that is expected.
- Billing follows cluster size, storage and egress traffic. You need to be able to read those three lines before committing, otherwise the surprise lands in month two.
Best for
- A product already in production whose slow queries trace back to the disk, not to a missing index.
- A team with someone who reads an infrastructure invoice and arbitrates a cluster size.
- A migration away from a managed Postgres where storage latency has become the measured bottleneck.