Google Cloud SQL Pricing: Estimate Compute, Storage, and HA

Cloud SQL pricing starts with the database engine, region, edition, machine shape, and time running. Enterprise and Enterprise Plus have different CPU and memory rates and capabilities, while SQL Server also adds license charges. For a dedicated-core instance, estimate monthly compute as provisioned vCPU-hours plus provisioned GiB-memory-hours; shared-core instances use instance pricing instead.

Do not size only from average CPU. Use measured peak connections, working-set memory, query latency, maintenance headroom, and growth. Cloud SQL bills a running instance even when application traffic is quiet, so an oversized always-on database can dominate an early product's backend bill. Compare one right-sized production shape with a peak-safe shape before choosing the edition.

Compute is only one line item. Add provisioned database storage, used backup storage, and any Hyperdisk Balanced capacity, IOPS, or throughput above its included baseline. Storage and network rates vary by region. If storage auto-increase is enabled, budget from expected retained data and growth rather than today's allocation alone.

High availability uses regional pricing, and failover or read-replica resources add material capacity. Read replicas are charged at the same rate as stand-alone instances. Also estimate cross-region replication and egress to application clients; keeping Cloud Run or Compute Engine clients in the same region can avoid some transfer charges and usually reduces latency.

Cloud SQL committed use discounts exchange a one- or three-year regional usage commitment for lower eligible CPU and memory prices. Apply them only to a stable baseline after measuring sustained demand. They do not reduce storage, network, instance, extended-support, or SQL Server license charges, so keep those items separate in the estimate.

Before launch, calculate a base month and a failure-tolerant month: primary compute, HA or replica capacity, storage growth, backups, network, and operational services such as logging and monitoring. Validate the selected region and engine against the current official Cloud SQL pricing page, then enter the wider application workload in the calculator.

Model Cloud SQL high availability, backup, and replica costs in detail.