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.
An organization is designing the disaster recovery architecture for an e-commerce application backed by Cloud SQL for MySQL. A business impact analysis establishes the following requirements:
RPO = 0) with automated failover and an RTO under 5 minutes.RPO under 5 minutes and an RTO under 15 minutes.Which architecture should the organization implement?
This architecture combines in-region High Availability (HA) with geo-redundant backups stored in a multi-region Cloud Storage location. HA protects against zonal failures through synchronous disk replication, while multi-region backups ensure a durable copy of the database exists outside the primary region.
RTO < 15 minutes: In a regional outage, there is no pre-provisioned standby instance running in a secondary region. An operator must create a new Cloud SQL instance, configure networking, and restore the entire dataset from backup, which frequently takes hours depending on database size.RPO < 5 minutes: Automated backups run on scheduled intervals (such as daily), meaning all transactions committed after the last completed backup and archived log transfer would be lost during a total regional failure.RPO = 0 and automated failover across zones.The application were a lower-criticality workload (such as an internal departmental tool) with lenient regional recovery targets, like an RTO of several hours or days and an RPO of up to 24 hours, where minimizing ongoing standby costs is the primary goal.
Backups remain essential alongside HA and read replicas to protect against logical data corruption, such as accidental table drops, because replication immediately propagates destructive SQL commands to replicas.
This configuration runs a single Cloud SQL primary instance in one zone without a local standby, paired with an asynchronous read replica in a different region for disaster recovery (DR). While it provides geographic redundancy across regions, it lacks High Availability (HA) within the primary region.
RPO = 0: Because cross-region replication is asynchronous, the primary commits transactions before they reach the remote replica. If the primary zone fails, any uncommitted transactions still in transit (replication lag) are lost or inaccessible until the zone recovers.The organization could tolerate non-zero data loss (RPO > 0) and manual cross-region intervention even during a single-zone outage, and prioritized minimizing compute costs over automated zonal resilience.
A common misconception is treating a read replica as a substitute for High Availability. HA protects against instance and zonal failures with synchronous replication and automatic failover inside a region, whereas cross-region replicas address regional disasters.
Combining Cloud SQL High Availability (HA) with a cross-region read replica creates a two-tier resilience architecture. Within the primary region, HA pairs a primary VM instance with a standby VM instance in a separate zone backed by a synchronously replicated Regional Persistent Disk. Across regions, a cross-region read replica receives continuous asynchronous replication from the primary instance to guard against full regional disasters.
RPO = 0, RTO < 5 minutes): Because the Regional Persistent Disk synchronously replicates every committed write across two zones, a zonal failure causes zero data loss. Cloud SQL automatically detects the outage, attaches the disk to the standby VM, and reroutes traffic without manual intervention.RPO < 5 minutes, RTO < 15 minutes): An operational cross-region read replica stays continuously updated in a secondary region, eliminating the lengthy delays of provisioning a new instance and restoring from backups. An administrator can manually promote the replica to become the new primary within minutes.RPO stays within target.Unlike backup-only strategies that suffer from multi-hour restoration times, or cross-region synchronous setups that degrade write latency, pairing in-region HA with an asynchronous cross-region replica directly aligns each failure domain with its exact RTO, RPO, and performance constraints.
This approach replaces managed Cloud SQL with self-managed MySQL running on Compute Engine VMs across two regions. In semisynchronous replication, the primary database waits to acknowledge each write commit to the client until at least one replica confirms receipt of the transaction.
The organization required strict self-managed control over the database engine on Compute Engine, could accept elevated write commit latency across regions, and deployed additional local standby nodes for zonal protection.
Even in self-managed MySQL deployments on Compute Engine, best practices recommend separating local HA (such as Hyperdisk Balanced High Availability across zones within a region for RPO = 0) from cross-region DR (using asynchronous replication to avoid write-latency bottlenecks).
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