Cloud Database Engineer Advanced Practice Exam: Hard Questions 2025

A global e-commerce platform is migrating to Google Cloud. They need a transactional database for orders with: (1) strong consistency for inventory updates, (2) multi-region high availability with automatic failover, (3) low-latency reads/writes for users in North America and Europe, and (4) the ability to run SQL with ACID semantics. They can tolerate higher write latency than a single-region setup but not application-managed conflict resolution. What should you recommend?

A fintech company runs Cloud SQL for PostgreSQL in a private IP configuration. They must meet these requirements: (1) regional HA, (2) automatic failover, (3) minimize data loss, (4) enforce that only workloads in specific GKE namespaces can connect, and (5) rotate credentials without redeploying workloads. Which design best meets the requirements with the least operational overhead?

A SaaS provider uses Cloud Spanner and needs to support per-tenant data residency: EU tenants’ data must remain in EU, US tenants in US. Some shared reference data can be global. Tenants can be added frequently and must not impact other tenants’ performance. The solution should minimize operational overhead and avoid running many separate projects. What is the best approach?

You manage a Cloud SQL for MySQL instance with private IP serving a latency-sensitive application. After a recent change, CPU is low but query latency intermittently spikes. Cloud Monitoring shows frequent waits on disk I/O and a growing history list length (or analogous indicator of long-running transactions). You suspect transaction and index issues. What is the most effective next step to diagnose with minimal production impact?

A company must provision Cloud SQL for PostgreSQL for a regulated workload. Requirements: (1) customer-managed encryption keys (CMEK), (2) private connectivity only, (3) automated backups with point-in-time recovery, (4) least-privilege admin model where DBAs can manage schemas but cannot modify instance-level network or encryption settings, and (5) separation of duties between security administrators and DBAs. Which setup best meets these requirements?

A high-throughput IoT ingestion system uses Bigtable. After adding new devices, throughput becomes uneven and some tablet servers are hot while others are underutilized. Row keys are currently formatted as <device_id>#<timestamp>. You cannot pause ingestion and must improve distribution without losing query capabilities for “latest readings per device” and time-range scans per device. What is the best change?

You are migrating a large, write-heavy PostgreSQL database (multi-TB) from on-prem to Cloud SQL for PostgreSQL with minimal downtime. Requirements: (1) continuous replication during cutover, (2) ability to validate data consistency, (3) avoid long-running locks during initial load, and (4) keep application writes flowing until a short final cutover window. What is the best migration approach?

A company is migrating from a sharded MySQL fleet to Cloud Spanner. The existing application relies on auto-increment integer primary keys and frequent cross-table joins. They also have a requirement to preserve write throughput and avoid hotspotting. Which schema/key design is most appropriate in Spanner?

During an Oracle-to-AlloyDB migration, the source uses sequences and triggers and relies on Oracle-specific SQL features. The application has a hard deadline and cannot be fully refactored before the first cutover. The team needs an approach that minimizes risk while enabling an incremental modernization plan. What should you recommend?

A mission-critical Cloud SQL for PostgreSQL (HA) database shows periodic connection storms and CPU spikes during failover tests. The application uses short-lived connections from many pods and does not reuse connections effectively. You must improve stability during failovers and reduce connection overhead without changing application code significantly. What is the best solution?