Cloud SQL Infrastructure
Infrastructure-level patterns for Google Cloud SQL PostgreSQL provisioning and operations.
Machine Type Selection
| Type | vCPU | RAM | Cost/mo | Use Case |
|---|---|---|---|---|
db-f1-micro | Shared | 0.6GB | ~$16 | Dev/test only (CPU throttling) |
db-g1-small | Shared | 1.7GB | ~$28 | Light staging workloads |
db-standard-1 | 1 | 3.75GB | ~$68 | Production-like staging |
db-standard-2 | 2 | 7.5GB | ~$136 | Small production |
db-standard-4 | 4 | 15GB | ~$272 | Medium production |
Selection Heuristics
| Concurrent Connections | Recommended Tier |
|---|---|
| <10 | db-f1-micro (dev only) |
| 10-30 | db-g1-small |
| 30-100 | db-standard-1 |
| 100-300 | db-standard-2 |
| 300+ | db-standard-4 or higher |
Storage Types
| Type | $/GB/mo | IOPS | Latency | Use Case |
|---|---|---|---|---|
PD_HDD | $0.09 | Lower | 10-20ms | Cost-sensitive staging |
PD_SSD | $0.17 | Higher | 1-5ms | Production, performance testing |
Rule: Use SSD for any workload where query latency matters.
High Availability
| Mode | Description | Cost | Use When |
|---|---|---|---|
ZONAL | Single zone | Base | Staging, cost-sensitive prod |
REGIONAL | Multi-zone failover | +35% | Production with SLA requirements |
Failover time: 60-120 seconds for REGIONAL HA.
Backup Strategies
| Environment | Retention | PITR | Rationale |
|---|---|---|---|
| Staging | 3 days | No | Cost savings, easy to recreate |
| Production | 30 days | Yes | Compliance, disaster recovery |
Quick Decision Matrix
| Scenario | Machine | Storage | HA | Monthly Cost |
|---|---|---|---|---|
| Dev/Test | db-f1-micro | 10GB HDD | ZONAL | ~$17 |
| Staging | db-g1-small | 20GB HDD | ZONAL | ~$30 |
| Prod (small) | db-standard-1 | 50GB SSD | REGIONAL | ~$110 |
| Prod (medium) | db-standard-2 | 100GB SSD | REGIONAL | ~$200 |
See reference.md for sizing calculations, VPC networking, and Terraform patterns.
See examples.md for complete Terraform configurations.