Cloud SQL — Managed relational databases
PostgreSQL, MySQL, or SQL Server run by Google: backups, high availability, upgrades, and patches handled for you.
Cloud SQL is a regular Postgres (or MySQL or SQL Server) database, but Google takes care of the ops side: daily backups, point-in-time recovery, version upgrades, machine patches, SSL certificates.
You pick the engine and version (Postgres 16, MySQL 8, ...), a machine size (vCPU + RAM), a disk size, a . Google gives you a connection string and credentials.
It connects to via a built-in secure proxy (Cloud SQL Auth Proxy) — no public IP needed, all traffic stays inside Google's network, authenticated via .
Enable High Availability: Google replicates the DB across two zones in the same region. If the primary fails, failover happens in ~60 seconds. You pay roughly 2x for that safety.
When NOT to use Cloud SQL: global sharding across continents (use Spanner), NoSQL / document needs (use Firestore), very high write throughput at scale (consider Bigtable or AlloyDB).
Comparison: Cloud SQL = AWS RDS = Azure SQL Database. All three: managed flavors of standard SQL engines with HA and auto-backups.
Grounded on https://cloud.google.com/sql/docs/introduction
Next up
Secret Manager — Store secrets safely
A vault for API keys, DB passwords, and JWT secrets. Versioned, IAM-controlled, mounted as env vars in Cloud Run or fetched at build time by Cloud Build.