GCP Professional Cloud Architect Practice Question 2: Minimal Downtime Database Migration
Google Cloud Professional Cloud Architect · Test 1 · question 2 of 10
During the third wave of a data-center exit, a retailer must move a 900 GB self-managed MySQL 8.0 database to Cloud SQL for MySQL. The application is a packaged storefront that cannot be modified, the database takes writes 24 hours a day, and the business will only approve a cutover window of about 15 minutes. A trial full export and import took just under nine hours.
What migration approach should the architect propose?
Show answer and explanation
Answer:
Database Migration Service supports both one-time and continuous migrations into Cloud SQL. A continuous job seeds the destination with a full copy and then keeps applying the source's ongoing changes, so the replication lag, not the size of the database, determines the cutover length. When the business is ready, you stop writes on the source, wait for lag to reach zero, and promote the destination to a standalone primary. That is how a nine-hour copy fits inside a fifteen-minute window, and it requires no application change because the target is MySQL-compatible.
Option A is the approach that was already measured and rejected. Dump and import is a one-time copy: every write taken after the dump starts is lost unless the source is frozen, which turns the nine hours into nine hours of downtime.
Option C moves the problem rather than solving it. Migrating the VM keeps the team responsible for patching, backups, replication, and failover, forfeiting the managed-service benefit that is the point of targeting Cloud SQL, and a VM migration still requires a cutover of its own.
Option D confuses an analytics pipeline with a transactional migration. Datastream into BigQuery is a change-data-capture path for analytics; BigQuery is not an OLTP engine and cannot serve the storefront's row-level reads and writes, and the packaged application cannot be rewritten anyway.
Reference: Types of migration (Database Migration Service for MySQL)