Managed Postgres on Fly.io

Managed Postgres is our fully managed database service. We run production PostgreSQL clusters for you and handle automatic backups and recovery, high availability with automatic failover, performance monitoring, resource scaling, and encryption of data at rest and in transit. You get a Postgres cluster inside your Fly.io organization's private network without doing the operational work yourself.

Run fly mpg create and pick an organization, a region, and a plan. You can also create a cluster from the Fly.io dashboard. Once it is running, fly mpg attach connects it to one of your apps by cluster ID and app name.

Attaching a cluster with fly mpg attach sets a DATABASE_URL secret on your Fly.io app. That URL points at the pooled endpoint, which routes through PgBouncer so a busy app does not exhaust Postgres connections. To open a psql session yourself, run fly mpg connect.

No. Your cluster runs inside your Fly.io organization's private network, so apps in the same organization reach it directly and nothing else can. From your own machine, connect through flyctl with fly mpg connect or fly mpg proxy, or over WireGuard.

Yes. Every plan runs a primary and a replica with automatic failover, and storage is replicated across all nodes in the cluster. High availability, automatic backups, and connection pooling are included on every plan rather than sold as add-ons.

Managed Postgres runs in Amsterdam, Dallas, Frankfurt, Sao Paulo, Ashburn, Los Angeles, London, Tokyo, Chicago, Singapore, San Jose, Sydney, and Toronto. We are adding regions as capacity comes online, so check the Fly.io dashboard for the current list when you create a cluster.

Yes. Every Managed Postgres image on Fly.io ships pgvector for vector similarity search, and you can turn on PostGIS for geospatial work when you create a cluster or later from the dashboard. A catalog of trusted PostgreSQL extensions sits alongside them, including pg_trgm, hstore, citext, and pgcrypto.

We back your cluster up automatically on a rolling schedule and keep those backups for ten days. fly mpg restore rebuilds a cluster from a backup or from a point in time into a new cluster on Fly.io, so recovering from a bad migration does not mean touching the original.

Yes. A cluster starts with one database named fly-db and an admin user named fly-user, and you can add more databases, schemas, and users on the same cluster. On Fly.io, fly mpg databases and fly mpg users manage both from the command line.

No. High availability is within one region: a primary and a replica in the cluster's region, with automatic failover between them. For a globally distributed app, keep the database in a single region, spread your Fly.io Machines across the regions your users are in, and cache in front of the database rather than trying to run a replica everywhere.