GCP Professional Data Engineer Practice Question

Dataform supports built-in data quality tests called assertions. An assertion is just a SQLX file whose SELECT statement returns the rows that violate a quality rule; when an assertion returns any rows, the Dataform run fails, downstream operations are skipped, and the failure is shown in the run logs and UI. Creating assertions that check for NULL customer_id values, duplicate order_id values, and negative revenue therefore stops the workflow automatically and keeps the implementation entirely inside the existing Dataform environment with minimal additional code.

Adding BigQuery NOT NULL or CHECK constraints would not stop a batch query in BigQuery today because those constraints are informational only. External Cloud Functions or Monitoring alerts would add custom components and would not natively cancel the Dataform run; they also introduce operational overhead and delayed feedback.