Cloud Database (RDS)

A managed database that just works — automatic primary/standby high availability, read replicas, and slow-query analysis. No hand-rolled backups or disaster recovery.

Core capabilities

Managed databases versus rolling your own

Run a database on your own server and you own version upgrades, backup scripts, replication, slow-query forensics, and failover. A managed database turns all of that into platform features, leaving you the schema and the queries.

The test is simple: if nobody on the team owns database operations, the managed version saves far more time than it costs in premium. If you need deep customization — exotic storage engines, unusual replication topologies — self-hosting is genuinely more flexible.

High availability does not cost what you expect

A primary/standby setup occupies two copies of storage and compute, so "HA doubles the bill" roughly holds — and the standby usually serves no read traffic.

A cheaper approach is tiering: give the transactional core HA, but run reporting and log databases as single instances with automated backups. Routing read-only traffic to replicas beats blindly upsizing the primary.

Storage and backups usually dominate the bill

The instance class is only part of the invoice. As data grows, storage, backup retention, and read replicas all climb. Pruning old backups and archiving cold data to object storage is the most direct way to hold this down.

Pairs well with

Frequently asked questions