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.
Your team is executing a minimal-downtime live migration from a production MySQL database to Spanner using Datastream, Cloud Storage, Pub/Sub, and a streaming Dataflow pipeline. Before the cutover window, you completed the initial backfill, confirmed the severe dead-letter queue (DLQ) directory is empty, and verified active Change Data Capture (CDC) replication. You have just paused all application write traffic to the source MySQL instance.
Which sequence of monitoring checks and pipeline actions should you perform next before redirecting application traffic to Spanner?
Canceling a streaming Dataflow job immediately terminates workers without waiting for in-flight bundles or Pub/Sub messages to finish processing, while the Data Validation Tool (DVT) compares schemas, row counts, and column aggregations between source and target databases.
0—risks leaving recently committed MySQL binary log events uncaptured.The goal were to abort a corrupted migration job that needed to be restarted from scratch, or to run post-migration data consistency checks after all replication pipelines had already drained cleanly.
During a live cutover, engineers must always drain rather than cancel streaming pipelines so that buffered change events are committed to Spanner before validating final consistency.
Database Migration Service (DMS) job promotion is the cutover step used in managed Cloud SQL or AlloyDB migrations to disconnect a target replica from its source, clean up database replication slots, and turn the destination into a standalone primary instance.
pglogical, pgnative, or SQL Server Distributed Availability Groups.You were migrating from PostgreSQL or SQL Server to Cloud SQL or AlloyDB using Database Migration Service, where waiting for replication lag to reach 0 and promoting the DMS migration job is the standard cutover procedure.
Recognizing which orchestration service manages your pipeline prevents mixing up DMS replica promotion with Datastream-to-Dataflow pipeline draining.
The Dataflow Replication lag total latency_MAX counter tracks the maximum delay of records currently flowing through the Dataflow pipeline, while Total permanent errors is a cumulative custom counter recording how many change events failed permanently during the job's lifetime.
Total permanent errors counter is cumulative and never decrements or resets to zero even after failed records are resolved and replayed.Total permanent errors remains greater than 0 for the life of the job once any permanent error occurs.severe/ and retry/ DLQ directories directly to verify an empty error state.system_latencies and freshness) must settle before downstream Dataflow metrics can be trusted.No such scenario makes this sequence valid; stopping Dataflow before upstream Datastream capture finishes and waiting for a cumulative counter to decrement are both operational anti-patterns.
Because Total permanent errors does not reset, checking the physical DLQ paths in Cloud Storage—both before pausing writes and after draining Dataflow—is mandatory for a clean cutover.
CDC cutover draining is the ordered verification and shutdown sequence used in an asynchronous Change Data Capture (CDC) pipeline to guarantee that every transaction committed before application writes were paused is safely applied to the target database. In a streaming architecture connecting MySQL to Spanner via Datastream, Cloud Storage, Pub/Sub, and Dataflow, data moves through multiple decoupled stages that each introduce buffering and latency.
datastream.googleapis.com/stream/freshness) and Throughput (datastream.googleapis.com/stream/event_count) metrics to drop to 0 confirms all binary log events up to the write pause have been captured before pausing the stream.Replication lag total latency_MAX, and Spanner Write throughput reach 0 verifies that no writes remain in transit.Stopping components abruptly or relying on a single lag metric risks cutting off records still buffered in Cloud Storage or Pub/Sub. This step-by-step metric verification and graceful drain sequence guarantees complete synchronization before redirecting application connections.
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