GCP Professional Cloud Architect Practice Question

You are migrating a mission-critical three-tier ordering platform from an on-premises data center to Google Cloud. Application servers will run on a regional managed instance group behind an external HTTP(S) load balancer, and the MySQL database will be moved to Cloud SQL using Database Migration Service (continuous replication). Business stakeholders demand an RPO of zero, a switchover window below five minutes, and evidence that the migrated stack can safely handle production load before the final cutover. Which test plan best meets these requirements?

After replication completes, power off the on-premises database, promote Cloud SQL to primary, run peak-load tests against the cloud stack, and roll back DNS only if problems are detected.
Before the migration window, perform a disaster-recovery failover test by simulating a regional outage in Google Cloud; if successful, immediately switch all user traffic to the cloud environment.
Keep replication running; execute smoke tests; configure the external HTTP(S) load balancer to route 1 percent of live traffic to the cloud back-end while monitoring errors; when metrics are clean, promote the Cloud SQL replica and shift traffic to 100 percent.
Freeze on-premises writes for four hours, export the database, import it into Cloud SQL, run synthetic load tests in the cloud, then repoint DNS to the new VIP.

Report Issue

Answer Description

Using continuous replication keeps the target Cloud SQL instance nearly in sync, eliminating data-loss risk. While replication continues, a low-risk canary can be performed by configuring weighted traffic splitting on the external HTTP(S) load balancer to send a small percentage of live production traffic to the new back-end while the majority still goes to the on-premises system. Smoke and functional checks run in parallel, and metrics can be monitored in real time. Once error rates and performance meet SLOs, the team can promote the Cloud SQL replica and shift the load balancer weight to 100 percent, requiring only the brief promotion window (<5 min) for full cutover. The other options either introduce hours of write downtime, perform load testing only after cutover, or test a DR scenario that does not validate live production traffic before migration, so they do not satisfy both the RPO and switchover objectives.

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.