GCP Professional Cloud Architect Practice Question

Your company operates a payments API on a single-zone GKE Autopilot cluster in us-central1 behind an external HTTP(S) load balancer. The PostgreSQL backend runs in a single-region Cloud SQL instance in the same region. An upcoming audit requires that the service remain available with an RTO of 5 minutes and an RPO of 15 minutes even if an entire Google Cloud region fails. The operations team has limited capacity to manage complex failover runbooks. Which architecture change best meets these business continuity requirements?

Create a cross-region read replica of the Cloud SQL instance in us-east1 and use scripted DNS changes to redirect traffic to a new zonal GKE cluster in that region after a failure occurs.
Enable high availability on the existing Cloud SQL instance and convert the single-zone GKE cluster to a regional cluster within us-central1 to achieve zone redundancy.
Migrate the database to Cloud Spanner with a multi-region configuration and deploy two GKE Autopilot clusters in separate regions, fronted by a global external HTTP(S) load balancer that uses multi-cluster Ingress for active-active traffic distribution.
Replace Cloud SQL with a self-managed PostgreSQL instance on Compute Engine and configure asynchronous replication to a standby VM in another region, triggering manual promotion during outages.

Report Issue

Answer Description

A multi-region Cloud Spanner instance keeps data synchronously replicated across at least three geographically separated zones in two or more regions, offering 99.999 % availability and automatic, lossless failover-well within a 5-minute RTO and 15-minute RPO. Deploying independent GKE Autopilot clusters in different regions and fronting them with a global external HTTP(S) Load Balancer that uses multi-cluster Ingress provides active-active serving and automatic traffic failover without manual intervention, satisfying the requirement for high availability and minimal operational overhead.

Enabling regional GKE or Cloud SQL High Availability within a single region only protects against zone failures, not region-wide outages. Cross-region read replicas or self-managed PostgreSQL on Compute Engine would still require manual promotion or operational runbooks, risking an RTO beyond 5 minutes and increasing maintenance burden. Therefore, the multi-region Cloud Spanner plus multi-region GKE deployment is the most appropriate choice.

Ask Bash

Bash is our AI bot, trained to help you pass your exam. AI Generated Content may display inaccurate information, always double-check anything important.