Maintaining and automating data workloads; security and compliance
Free Google Cloud Certified - Professional Data Engineer practice — 6 questions on Maintaining and automating data workloads; security and compliance, with explanations. No sign-up. Full 12-question mixed test →
A Cloud Composer environment hosts an hourly DAG with catchup=True (the Airflow default). The environment underwent unplanned maintenance and was offline for 30 hours. On restart, the engineer wants Airflow to execute only the single most recent scheduled interval, not all 30 missed intervals. Which change achieves this with minimal side effects?
catchup is a per-DAG setting that controls whether the scheduler backfills every missed schedule interval since the last successful run. Setting catchup=False tells the scheduler to skip missed intervals and only create a DAG run for the most recent schedule, exactly matching the requirement.
A streaming Dataflow pipeline reads from Pub/Sub, performs windowed aggregations, and writes to BigQuery. Engineers need an early warning when the pipeline is falling behind incoming events, before downstream dashboards show stale results. Which Cloud Monitoring metric should the alerting policy be based on?
Data freshness (job/data_watermark_age) measures the difference between the pipeline's output watermark and real time. A growing value directly indicates the pipeline cannot keep pace with incoming events, which is exactly the 'falling behind' condition being monitored.
A company's BigQuery dataset is created in the single region us-central1. Compliance requires that in a regional outage the dataset can be restored elsewhere with a recovery point objective (RPO) of 24 hours. What should be implemented?
A dataset created in a single region (us-central1) is not automatically replicated across regions. Scheduling a BigQuery Data Transfer Service dataset copy job on a daily cadence to a dataset in another region gives a concrete, verifiable RPO of 24 hours and a target location to restore from during an outage.
Data analysts need to author and run BigQuery scheduled queries that write results into a destination dataset, but must not be able to delete the dataset or modify its IAM policy. Which role assignment best satisfies least privilege?
roles/bigquery.dataEditor on the destination dataset grants the table/data-level permissions needed to write scheduled query results, while roles/bigquery.jobUser at the project level allows running query jobs. Neither role grants dataset-level setIamPolicy or dataset-delete permissions, satisfying least privilege.
A Dataflow job runs inside a VPC Service Controls perimeter that also protects several BigQuery datasets. The job must consume messages from a Pub/Sub topic that lives in a separate project not included in the perimeter, and that project cannot be added to the perimeter for organizational reasons. How should access be granted while keeping the perimeter's protections intact?
VPC Service Controls supports egress rules that identify a specific identity (the Dataflow service account) and a specific external resource (the Pub/Sub topic/project) that is permitted to be accessed from inside the perimeter, without adding the whole external project to the perimeter or disabling protection.
A BigQuery table contains a national ID column. Regular analysts should see the ID with only the last 4 digits exposed, while the compliance team must see the full unmasked value for audits. Both groups need to query the same live table without maintaining separate copies. Which approach meets this requirement?
BigQuery column-level security with Data Catalog policy tags supports dynamic data masking: the same underlying column is masked at query time for principals with the Masked Reader role, while principals granted Fine-Grained Reader (or Data Policy admin equivalent unmasked access) see the full value -- all against one live table with no duplication.
Ready for the real thing?
The full course: two full-length practice tests, video lessons for every exam domain, hands-on labs and detailed explanations.
$129.99 $34.99 with code FREETEST33 — valid through September 16.
Get my $34.99 deal →