A realistic eight-week plan at roughly 10 hours per week. Build everything hands-on in a real project using the free tier and the free BigQuery sandbox. Compress to six weeks if you already have GCP data experience; extend if you are new to it.
| Week | Focus | Checkpoint |
|---|---|---|
| 1 | Storage choices: BigQuery vs Bigtable vs Cloud Storage vs Cloud SQL/Spanner | You can match a scenario to the right store and say why |
| 2 | BigQuery basics: loading, SQL, schemas; partitioning and clustering | You can load data and tune a table for cost and speed |
| 3 | Ingestion: Pub/Sub, batch loads, streaming vs batch | You can ingest a stream and a batch into BigQuery |
| 4 | Dataflow (Apache Beam): batch and streaming pipelines, windowing | You can build a working Dataflow pipeline |
| 5 | Dataproc vs Dataflow; orchestration with Cloud Composer (Airflow) | You can choose between them and schedule a pipeline |
| 6 | Governance and security: IAM, Dataplex, data quality and discovery | You can apply least-privilege IAM and govern a dataset |
| 7 | Operations: monitoring, logging, troubleshooting, automating workloads | You can monitor a pipeline and recover from a failure |
| 8 | Official sample questions and full-length timed reviews | You are consistently correct across all five domains |
Final tips
Domains 2 and 3 (ingesting/processing and storing) are the largest and most service-heavy - give them the most time, and centre your study on when to choose each service. Learn BigQuery cost (partitioning, clustering, bytes scanned), not just SQL syntax. Governance and operations span two domains worth more than a third of the exam, so do not skip IAM, Dataplex and monitoring. Because Google does not publish a passing score, aim to be comfortably correct across all five domains on fresh questions before booking. Avoid “exam dump” sites - they breach Google’s certification policy and teach the wrong habits.