Prepare and test your skills

Prepare and test your skills

Free sample question with answers and detailed explanation below. The Professional Cloud Database Engineer practice bank has an additional 769 practice questions.Try 10 free realistic questions: instant score, no credit card, and no account required.
A financial services company is migrating a mission-critical PostgreSQL database from Amazon RDS to Cloud SQL for PostgreSQL. The engineering team has established the following migration and contingency requirements:
Which fallback strategy should the team implement?
A simple fallback is a reactive rollback strategy in which applications are switched back to the original source database after a failed cutover. When post-cutover transactions on the target database must be preserved, the team stops writes to the target (Cloud SQL), takes a database backup, restores that backup onto the original source (Amazon RDS), and reconnects the applications.
The workload could tolerate significant scheduled downtime during both migration and rollback, or if the business did not need to retain transactions committed on the target database after cutover.
While a simple fallback is adequate for non-critical workloads or scheduled maintenance windows, mission-critical transactional systems requiring zero downtime and zero data loss must use duplicate writes or continuous replication mechanisms instead.
The duplicate-writes strategy (also known as a dual-write or Y-migration architecture) is an application-level migration and fallback pattern in which an application writes every data modification to both the legacy source database and the new target database at the same time. When all services interact with storage through a centralized data-access microservice, the dual-write logic is implemented in one proxy layer rather than across every individual application service.
Unlike reverse replication, which leaves a brief window without fallback after cutover and cannot be tested end-to-end beforehand, duplicate writes through a data-access microservice provide continuous, zero-downtime rollback protection and full support for a gradual cutover.
Forward replication is a variation of reverse replication in which writes committed on the newly promoted target database (Cloud SQL) are continuously replicated to a third database instance rather than back to the original source database. Applications can be pointed to this third database to run read-only queries or maintain availability if the primary target database becomes unavailable.
The organization needed a fallback mechanism that could be fully tested end-to-end before cutover, was required to decommission the original Amazon RDS instance immediately after cutover, and only needed applications to execute read-only queries against a standby instance during an outage.
Database architects sometimes confuse forward replication with duplicate writes. Forward replication uses database-level replication downstream from the target to a third environment, whereas duplicate writes use application-level logic to keep both the original source and target synchronized simultaneously.
Reverse replication is a database-level fallback strategy used with continuous replication migrations. After production cutover redirects application writes to the new target database (Cloud SQL), a replication pipeline captures those changes—typically via change data capture (CDC) or transaction logs—and replicates them back to the original source database (Amazon RDS) to keep it synchronized.
The team had no capacity to refactor application or microservice code, wanted an operations-driven migration with minimal development effort, and could tolerate brief cutover and fallback downtime windows while accepting a short post-cutover period without pre-tested fallback.
While reverse replication is a common fallback pattern for managed migrations (such as change stream pipelines or failback jobs), it cannot match the zero-downtime, pre-validated gradual cutover capabilities of a duplicate-writes architecture.
Answer them yourself and get your score the moment you finish. No account needed.
Full access includes
Keep the momentum going with these hand-picked practice scenarios
Last updated Validated