Google Cloud 上でのスケーラブル・高可用 DB のデプロイ

本番推奨設定

resource "google_sql_database_instance" "main" {
  name             = "my-postgres-prod"
  database_version = "POSTGRES_15"
  region           = "asia-northeast1"

  settings {
    tier              = "db-custom-4-15360"
    availability_type = "REGIONAL"   # ← HA

    backup_configuration {
      enabled                        = true
      start_time                     = "02:00"
      point_in_time_recovery_enabled = true
      transaction_log_retention_days = 7
    }

    ip_configuration {
      ipv4_enabled    = false              # ← Public IP 無効
      private_network = google_compute_network.main.id
    }

    insights_config {
      query_insights_enabled = true
    }
  }

  deletion_protection = true             # ← 誤削除防止
}

Cloud SQL HA フェイルオーバー試験コマンド

# 手動フェイルオーバー
gcloud sql instances failover my-mysql-prod

# 動作確認
# 1. アプリから接続できるか
# 2. 書き込みができるか
# 3. データ整合性が保たれているか
# 4. フェイルオーバー時間 (数十秒〜数分) を計測

GitOps パターン

Cloud Monitoring Alert を Terraform で

resource "google_monitoring_alert_policy" "cpu" {
  display_name = "Cloud SQL CPU High"
  conditions {
    display_name = "CPU > 80% for 5 min"
    condition_threshold {
      filter          = "resource.type=\"cloudsql_database\" AND ..."
      duration        = "300s"
      comparison      = "COMPARISON_GT"
      threshold_value = 0.8
    }
  }
  notification_channels = [google_monitoring_notification_channel.slack.id]
}