GCP Professional Data Engineer Practice Question

A fintech company runs a 40-TB Oracle 11g data warehouse on-premises. About 800 GB of new and updated data is generated every week. Business stakeholders require a migration to BigQuery with two hard constraints:

Cutover downtime must not exceed 30 minutes, because analysts run near-real-time reports during trading hours.
Before the switch, data engineers must be able to run reconciliation queries in BigQuery while the on-premises system continues serving production workloads.

You have already established a 10-Gbps Cloud Interconnect link between the datacenter and Google Cloud, and VPC-SC boundaries are in place. Which migration plan best meets the requirements while minimizing operational overhead?

Use Datastream to capture Oracle change data and land it in Cloud Storage, configure BigQuery external tables or frequent loads for validation, then schedule a brief cutover once the change stream is caught up.
Develop a custom Apache Beam CDC pipeline on Dataflow that streams Oracle redo logs into BigQuery in real time, allowing validation before triggering DNS cutover.
Ship initial data on a Transfer Appliance, then perform daily incremental loads with Storage Transfer Service and a final rsync-based cutover script.
Set up Database Migration Service in continuous-replication mode from Oracle to BigQuery and cut over after replication lag reaches zero.

Report Issue

Answer Description

Datastream can continuously capture change data (CDC) from on-premises Oracle and write it in near real-time to Cloud Storage, from which BigQuery can ingest the files with low latency using external tables or scheduled loads. This lets the team validate data in BigQuery while production continues on Oracle. At cutover, you only need a brief pause to let the CDC queue drain and switch reporting to BigQuery, keeping downtime within 30 minutes. Transfer Appliance is optimized for bulk, offline moves and would not satisfy the near-zero-downtime or ongoing validation requirement. Database Migration Service does not yet support Oracle sources for continuous migration, and building a custom Dataflow CDC pipeline would meet the SLOs but carries higher development and maintenance overhead than the managed Datastream service.

Ask Bash

Bash is our AI bot, trained to help you pass your exam. AI Generated Content may display inaccurate information, always double-check anything important.