Prepare and test your skills

Prepare and test your skills

Free sample question with answers and detailed explanation below. The Professional Cloud Developer practice bank has an additional 608 practice questions.Try 10 free realistic questions: instant score, no credit card, and no account required.
You are designing a globally distributed financial trading platform that requires a relational database. Your application has the following requirements:
Which database architecture should you implement?
AlloyDB Omni with Patroni enables PostgreSQL-compatible clusters to use a mixed replication topology: synchronous streaming replication between a primary node and a nearby standby within the same region, combined with asynchronous streaming replication to standby nodes in distant regions for disaster recovery.
etcd) and load balancers (such as HAProxy) that add operational complexity compared to fully managed services.max_standby_streaming_delay or query routing is carefully tuned.This architecture is ideal when you need low write latency in a primary region with RPO = 0 for zonal failures, while accepting potential data loss and stale regional reads for cross-region disaster recovery or hybrid/on-premises deployments.
Standard PostgreSQL streaming replication forces a strict trade-off between cross-region write latency (synchronous) and data loss/staleness (asynchronous), whereas Spanner's Paxos quorum and TrueTime bridge geographic distance with external consistency.
AlloyDB for PostgreSQL separates compute and storage, allowing a primary instance and multiple read pool instances (up to 20 nodes per cluster) to share a distributed storage layer while replicating Write-Ahead Log (WAL) records to keep replica caches updated. The flag google_storage.log_replay_throttle_read_transactions controls whether AlloyDB pauses new read queries when replication lag exceeds one second.
google_storage.log_replay_throttle_read_transactions to off explicitly prioritizes read query latency over data freshness, allowing replicas to serve increasingly stale data instead of pausing to let WAL replay catch up.google_storage.log_replay_throttle_read_transactions set to on reduces staleness at the cost of query delays up to one minute during heavy write bursts.This setup works well for single-region Hybrid Transactional and Analytical Processing (HTAP) or read-heavy workloads where low read query latency is more important than reading the latest committed state.
Students often confuse AlloyDB's shared storage with zero-lag reads. Because read pool nodes maintain independent memory caches, WAL replay lag still makes read pools eventually consistent relative to the primary.
Cloud Spanner is a globally distributed, horizontally scalable relational database that provides external consistency—the strictest consistency guarantee for transaction-processing systems. It achieves this across multiple regions using TrueTime, a distributed clock API that assigns globally ordered, monotonically increasing timestamps, combined with Multi-Version Concurrency Control (MVCC) and Paxos-based synchronous replication.
Unlike Cloud SQL and AlloyDB, which rely on asynchronous replication for cross-region distribution and introduce replication lag, Spanner natively combines relational ACID semantics, horizontal read/write scaling, and synchronous multi-region durability.
Cloud SQL is a fully managed relational database service for MySQL, PostgreSQL, and SQL Server. Its High Availability (HA) configuration uses a primary and standby VM in two zones within a single region backed by a Regional Persistent Disk that is synchronously replicated across those two zones. To scale reads or protect against regional outages, Cloud SQL uses cross-region read replicas.
replica_lag and network_lag.This architecture is appropriate when an application operates primarily in one region, can tolerate eventual consistency on cross-region read replicas, and accepts a non-zero RPO during rare regional disasters.
While increasing replica size or enabling parallel replication reduces lag, asynchronous replication cannot eliminate the physics of cross-region round-trip delays without switching to a synchronous consensus engine like Cloud Spanner.
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