GCP Study Hub

Databases

Cloud SQL

Fully managed relational database service

AWS equivalent

RDS

RelationalMySQLPostgreSQLSQL Server
🔄

AWS → GCP: Key Differences

  • Very similar to RDS. Supports MySQL, PostgreSQL, and SQL Server.

  • High availability: standby replica in another zone with automatic failover. ≈ RDS Multi-AZ.

  • Cloud SQL Auth Proxy: secure connection to Cloud SQL without managing SSL certs or whitelisting IPs.

  • No Oracle support. For Oracle migrations, use Bare Metal Solution or consider refactoring.

📌

Key Concepts to Know

  • 1

    Automated backups, point-in-time recovery (PITR) — same concepts as RDS.

  • 2

    Storage auto-increase: Cloud SQL can automatically grow storage. Doesn't auto-shrink.

  • 3

    Private IP: connect from within VPC without public internet.

  • 4

    Database Migration Service: migrate from on-prem MySQL/PostgreSQL to Cloud SQL with minimal downtime using CDC.

💡

DCE Interview Tips

  • For Thai banks migrating from on-prem MySQL: 'Cloud SQL is a direct lift with minimal changes. Use Database Migration Service for online migration with near-zero downtime.'

  • HA is not enabled by default — important to mention for production workloads.

  • Cloud SQL vs Spanner: 'Cloud SQL for existing relational workloads under ~10TB. Spanner when you need horizontal scaling and global consistency.'

⚠️

Common Gotchas

  • !

    Cloud SQL is zonal by default. You must explicitly enable HA for production — this doubles the cost.

  • !

    Cloud SQL instances don't survive region failures — only zone failures with HA. For true multi-region, consider Spanner.

  • !

    Max storage: 64TB. Beyond that, you need Spanner or BigQuery.