Download [PDE] Google Cloud Certified Professional Data Engineer Certification Exam Guide and more Exams Technology in PDF only on Docsity!
Data Engineer Certification Exam Guide
Question 1. Which Google Cloud service is primarily used to enforce fine‑grained access control over BigQuery datasets? A) Cloud IAM B) Cloud KMS C) Cloud DLP D) Cloud Asset Inventory Answer: A Explanation: Cloud IAM lets you assign roles and permissions at the dataset level, enabling granular access control for BigQuery. Question 2. In a data‑processing pipeline, which Google Cloud component provides serverless stream processing with Apache Beam semantics? A) Cloud Dataflow B) Cloud Dataproc C) Cloud Composer D) Cloud Run Answer: A Explanation: Cloud Dataflow runs Apache Beam pipelines in a fully managed, serverless environment for both batch and streaming workloads. Question 3. Which encryption method protects data at rest in Google Cloud Storage by default? A) Customer‑Supplied Encryption Keys (CSEK) B) Customer‑Managed Encryption Keys (CMEK) C) Google‑Managed Encryption Keys (GMEK) D) Transparent Data Encryption (TDE)
Data Engineer Certification Exam Guide
Answer: C Explanation: GMEK is automatically applied to all Cloud Storage objects, providing encryption at rest without user intervention. Question 4. When designing a globally consistent relational database on GCP, which service should you choose? A) Cloud SQL B) Cloud Spanner C) AlloyDB D) Bigtable Answer: B Explanation: Cloud Spanner offers horizontal scalability with strong, external consistency across regions. Question 5. Which Google Cloud service is best suited for low‑latency, high‑throughput time‑series data from IoT devices? A) Cloud Firestore B) Cloud Bigtable C) Cloud SQL D) Cloud Datastore Answer: B Explanation: Cloud Bigtable is optimized for massive write throughput and low‑latency reads, ideal for time‑series and IoT workloads. Question 6. To automatically delete objects older than 365 days in a Cloud Storage bucket, you should configure a: A) Object versioning policy
Data Engineer Certification Exam Guide
Question 9. Which Google Cloud service provides a managed Apache Airflow environment for orchestrating data workflows? A) Cloud Composer B) Cloud Scheduler C) Cloud Functions D) Cloud Run Answer: A Explanation: Cloud Composer is a fully managed Airflow service that schedules and monitors complex pipelines. Question 10. When you need to enforce GDPR‑compliant data residency for a dataset stored in BigQuery, you should: A) Use a multi‑regional dataset in us‑central1 B) Store data in a regional dataset within the EU location C) Enable Cloud DLP to mask data D) Replicate data to asia‑north1 Answer: B Explanation: Placing the dataset in an EU regional location ensures data never leaves the EU, satisfying GDPR residency requirements. Question 11. Which Google Cloud product enables you to discover, profile, and tag data assets across multiple clouds? A) Data Catalog B) Dataplex C) Dataform
Data Engineer Certification Exam Guide
D) Data Fusion Answer: B Explanation: Dataplex provides data governance, discovery, and metadata management across GCP, on‑prem, and other clouds. Question 12. To encrypt data in transit between a client application and Cloud Pub/Sub, you must use: A) TLS 1. B) SSL only C) TLS 1.2 or higher D) No encryption, Pub/Sub is already secure Answer: C Explanation: Pub/Sub requires TLS 1.2+ for all client‑to‑service communication, ensuring data is encrypted in transit. Question 13. Which BigQuery storage format enables querying data stored directly in Cloud Storage without loading it? A) BigQuery Data Transfer Service B) BigLake (Lakehouse) tables C) External tables via Cloud SQL D) Legacy SQL tables Answer: B Explanation: BigLake allows you to create external tables that query Parquet, ORC, or Avro files in GCS as if they were native BigQuery tables. Question 14. When designing a data lake on GCP, which storage class is most cost‑effective for infrequently accessed archival data?
Data Engineer Certification Exam Guide
Explanation: Cloud Source Repositories (or any Git repository) allow you to store and version SQL scripts, enabling automated deployments. Question 17. Which BigQuery feature helps reduce query cost by caching results for identical queries within 24 hours? A) Slot reservations B) Query caching C) Auto‑materialized views D) Dataflow streaming inserts Answer: B Explanation: BigQuery automatically caches query results, and subsequent identical queries retrieve the cached data without re‑processing. Question 18. For a data warehouse workload that requires sub‑second latency dashboards, which BigQuery optimization should you prioritize? A) Partitioning by ingestion time B) Clustering on frequently filtered columns C) Using Standard SQL over Legacy SQL D) Increasing the number of slots via reservations Answer: D Explanation: Reserving more slots provides dedicated processing capacity, reducing latency for high‑concurrency dashboard queries. Question 19. Which GCP service provides a managed, serverless environment for running containerized AI models? A) AI Platform Prediction (now Vertex AI Prediction) B) Cloud Run
Data Engineer Certification Exam Guide
C) App Engine D) Cloud Functions Answer: A Explanation: Vertex AI Prediction serves trained models in a fully managed, autoscaling environment, supporting containerized model deployments. Question 20. To mask personally identifiable information (PII) in a dataset before loading into BigQuery, you would use: A) Cloud IAM policies B) Cloud DLP API C) Cloud KMS D) BigQuery Row‑level security Answer: B Explanation: Cloud DLP provides built‑in transformations like masking, tokenization, and redaction for PII. Question 21. Which windowing type in Apache Beam creates windows that close after a period of inactivity? A) Fixed windows B) Sliding windows C) Session windows D) Global windows Answer: C Explanation: Session windows dynamically group events based on gaps of inactivity, closing when no new events arrive within the gap duration.
Data Engineer Certification Exam Guide
Answer: C Explanation: The HA (regional) configuration provisions a standby instance in a different zone, providing automatic failover. Question 25. Which BigQuery storage option is most appropriate for a table that is updated daily with new partitions? A) Unpartitioned table B) Date‑sharded tables C) Partitioned tables on a DATE column D) Clustering only Answer: C Explanation: Partitioned tables on a DATE column allow efficient addition of daily partitions and improve query pruning. Question 26. When integrating Vertex AI with BigQuery ML, which statement is true? A) Vertex AI can directly train models on raw GCS files only. B) BigQuery ML models can be exported to Vertex AI for hyperparameter tuning. C) Vertex AI replaces BigQuery ML entirely. D) BigQuery ML cannot use data stored in BigLake tables. Answer: B Explanation: You can export a trained BigQuery ML model to Vertex AI for advanced training, tuning, and deployment. Question 27. Which Google Cloud service provides a unified view of logs, metrics, and traces for data pipelines? A) Cloud Logging
Data Engineer Certification Exam Guide
B) Cloud Monitoring C) Cloud Trace D) Cloud Operations Suite (formerly Stackdriver) Answer: D Explanation: Cloud Operations Suite integrates Logging, Monitoring, and Trace, giving a consolidated observability platform. Question 28. In a multi‑cloud analytics scenario using BigQuery Omni, data resides in: A) Only Google Cloud Storage B) Only BigQuery native tables C) External data warehouses like Snowflake or Azure Synapse D) On‑prem PostgreSQL only Answer: C Explanation: BigQuery Omni lets you query data stored in external cloud warehouses (AWS, Azure) without moving it. Question 29. Which of the following is a best practice for designing a data mesh on GCP? A) Centralize all data in a single BigQuery dataset. B) Use domain‑owned data products with self‑service APIs. C] Replicate every dataset across all regions. D) Store all raw data in Cloud SQL. Answer: B Explanation: Data mesh emphasizes decentralized, domain‑driven data ownership and self‑service interfaces.
Data Engineer Certification Exam Guide
Answer: A Explanation: Cloud Functions offers a serverless, pay‑per‑invocation environment ideal for lightweight ETL scripts. Question 33. Which Cloud IAM role grants read‑only access to all datasets in a project? A) roles/bigquery.dataEditor B) roles/bigquery.dataOwner C) roles/bigquery.dataViewer D) roles/bigquery.jobUser Answer: C Explanation: The bigquery.dataViewer role provides read‑only permissions on all datasets within a project. Question 34. To reduce network egress costs when moving data between GCP regions, you should: A) Use Cloud Interconnect B) Enable VPC‑Peering across regions C) Store data in Multi‑Regional buckets only D) Use Cloud CDN Answer: A Explanation: Dedicated Interconnect or Partner Interconnect offers lower‑cost, high‑throughput connectivity between regions. Question 35. Which of the following statements about Cloud Spanner’s “strong consistency” is correct? A) Reads are always served from the nearest replica, possibly stale.
Data Engineer Certification Exam Guide
B) Writes are eventually consistent across regions. C) Reads are guaranteed to see the most recent committed write. D) Consistency can be toggled per table. Answer: C Explanation: Cloud Spanner provides external strong consistency, ensuring reads reflect the latest committed transaction. Question 36. When using Pub/Sub with Dataflow for exactly‑once processing, which feature must be enabled? A) Message ordering B) Dead‑letter topics C) Dataflow streaming engine with Pub/Sub Lite D) Pub/Sub Exactly‑once delivery (via acknowledgments) Answer: D Explanation: Pub/Sub’s exactly‑once delivery ensures that each message is processed only once when combined with Dataflow’s checkpointing. Question 37. Which option best describes the purpose of a “materialized view” in BigQuery? A) To store raw data before transformation B) To automatically refresh a pre‑computed query result set C) To enforce column‑level security D) To replace partitioned tables Answer: B Explanation: Materialized views maintain an up‑to‑date copy of a query’s results, improving query performance.
Data Engineer Certification Exam Guide
D) Cloud Armor security policies Answer: C Explanation: IAM conditions allow you to restrict access based on attributes like user email domain. Question 41. Which of the following is NOT a valid BigQuery table partitioning type? A) Ingestion‑time partitioning B) Date‑column partitioning C) Integer‑range partitioning D) String‑hash partitioning Answer: D Explanation: BigQuery supports ingestion‑time, column‑based (DATE/TIMESTAMP), and integer‑range partitions, but not string‑hash partitioning. Question 42. When you need to perform ad‑hoc data profiling on GCS files, which service provides built‑in profiling capabilities? A) Dataform B) Dataplex C) Data Fusion D) Cloud Datalab Answer: B Explanation: Dataplex can scan, profile, and catalog data assets stored in Cloud Storage. Question 43. Which feature of Cloud Composer helps you visualize DAG dependencies? A) Airflow UI’s Graph view B) Cloud Scheduler dashboard
Data Engineer Certification Exam Guide
C) Cloud Logging logs explorer D) Cloud Monitoring dashboards Answer: A Explanation: The Airflow UI’s Graph view displays task dependencies within a DAG. Question 44. To minimize latency for a dashboard that queries recent 24‑hour data, you should: A) Use a table partitioned by month B) Use a table clustered on the timestamp column C) Store data in Cloud SQL D) Disable query caching Answer: B Explanation: Clustering on the timestamp column improves pruning for recent‑time queries, reducing latency. Question 45. Which GCP service can be used to orchestrate serverless functions as part of a data workflow without managing an Airflow environment? A) Cloud Composer B) Cloud Workflows C) Cloud Scheduler D) Cloud Build Answer: B Explanation: Cloud Workflows enables you to stitch together serverless services like Cloud Functions and Cloud Run in a controlled sequence.
Data Engineer Certification Exam Guide
Answer: C Explanation: Setting a Customer‑Managed Encryption Key (CMEK) as the default encryption for a bucket forces all objects to be encrypted with that key. Question 49. Which method allows you to run a Spark job on GCP without provisioning a cluster, while still using the full Spark API? A) Cloud Dataflow B) Dataproc Serverless C) Cloud Run D) Cloud Functions Answer: B Explanation: Dataproc Serverless runs Spark jobs in a managed environment without the need for a persistent cluster. Question 50. In BigQuery, what does the “slot” resource represent? A) A storage unit for table data B) A compute capacity unit for query execution C) A user permission role D) A network bandwidth allocation Answer: B Explanation: Slots are units of processing capacity that execute SQL queries; reservations allocate a fixed number of slots. Question 51. Which service should you use to create a unified, searchable catalog for datasets across multiple GCP projects? A) Cloud Asset Inventory
Data Engineer Certification Exam Guide
B) Data Catalog C) Cloud Search D) Dataplex Answer: B Explanation: Data Catalog provides a centralized metadata repository with search capabilities across projects. Question 52. To automatically retry failed Dataflow streaming jobs, you configure: A) Pub/Sub dead‑letter topics B) Dataflow job’s “max‑workers” setting C) Dataflow’s “autoscaling algorithm” with “worker‑restart” policy D) Cloud Scheduler cron jobs Answer: C Explanation: The autoscaling algorithm can be set to restart workers on failure, ensuring continuous streaming processing. Question 53. Which of the following is a valid reason to choose Cloud Firestore over Cloud Bigtable for a mobile app backend? A) Need for massive write throughput of billions of rows per second B) Requirement for strong consistency across continents C) Need for hierarchical data modeling with real‑time sync D) Requirement for columnar storage and analytical queries Answer: C Explanation: Cloud Firestore provides document‑oriented storage with real‑time synchronization, ideal for mobile app use cases.
