Google Cloud · PDE
Google Cloud Certified - Professional Data Engineer (PDE) Practice Tests
Google Cloud Professional Data Engineer practice aligned to the current v4.2 design, ingestion, storage, analysis, automation, and operations guide.
Exam Details
Practice Questions
1,063
≈ 21 practice exams
Duration
120 minutes
Passing Score
Not publicly disclosed
Difficulty
ProfessionalLast Updated
Oct 2026
Topics Covered
Design Data Processing Systems (22%)Ingest and Process Data (25%)Store Data (20%)Prepare and Use Data for Analysis (15%)Maintain and Automate Data Workloads (18%)
Google Cloud Professional Data Engineer Practice Exam
Google's v4.2 Professional Data Engineer guide covers designing data processing systems (about 22%), ingesting and processing data (25%), storing data (20%), preparing and using data for analysis (15%), and maintaining and automating workloads (18%). Current objectives include AI data enrichment, embeddings and retrieval-augmented generation alongside BigQuery, Dataflow, Pub/Sub, Dataproc, Dataplex, and Composer.
The standard exam has 40-50 multiple-choice and multiple-select questions in two hours and costs $200 USD plus tax. It is available in English and Japanese online or at a testing center. Google publishes no passing score, requires no prerequisite, recommends three years of industry experience including one year on Google Cloud, and makes the credential valid for two years.
Use these 1,063 questions to practise choosing a design from requirements for security, reliability, latency, portability, governance, and cost. Renewal holders can choose the standard exam, a one-hour $100 renewal exam with 20 questions, or eligible Google Skills coursework; always check your certification dashboard for eligibility before choosing a path.
Exam Overview
Google Cloud Professional Data Engineer validates design, construction, deployment, monitoring, maintenance, optimisation, security, and governance of data workloads. Version 4.2 adds current service and AI contexts such as Dataform, Dataplex Catalog, AI data enrichment, embeddings, retrieval-augmented generation, and LLM-assisted query generation while retaining core data-platform architecture.
Who Should Take This Exam
Google recommends three or more years of industry experience, including at least one year designing and managing solutions on Google Cloud. The credential targets data engineers and architects who can select services and patterns from workload, business, security, regulatory, reliability, performance, and cost requirements.
Prerequisites
There is no formal prerequisite. Candidates need working knowledge of SQL, batch and streaming processing, data modelling and storage, governance, IAM, encryption, observability, orchestration, data quality, and the principal Google Cloud data services.
Exam Format
The standard exam contains 40-50 multiple-choice and multiple-select questions in two hours. It costs $200 USD plus applicable tax, is available in English and Japanese online or at a test center, and has no publicly disclosed passing score. The credential is valid for two years.
Skills Measured
- 1.Design Data Processing Systems (about 22%): security, compliance, reliability, fidelity, portability, governance, and migration.
- 2.Ingest and Process Data (about 25%): batch and streaming pipelines, transformations, AI enrichment, orchestration, networking, encryption, and CI/CD.
- 3.Store Data (about 20%): service selection, warehouses, lakes, models, access patterns, lifecycle, performance, cost, and federated governance.
- 4.Prepare and Use Data for Analysis (about 15%): BI, query performance, masking, ML preparation, embeddings, RAG, and data sharing.
- 5.Maintain and Automate Data Workloads (about 18%): cost and capacity, orchestration, monitoring, troubleshooting, fault tolerance, replication, and failover.
Study Tips
- Use the current v4.2 official exam guide and map every bullet to a service, pattern, or hands-on task.
- Practise BigQuery design, optimisation, security, sharing, reservations, and operations across multiple scenarios.
- Build batch and streaming pipelines with Pub/Sub, Dataflow, Dataproc, storage, orchestration, monitoring, and recovery.
- Study current AI-data objectives including enrichment, embeddings, RAG preparation, and LLM-assisted query work.
- Choose answers from explicit business, security, performance, reliability, and cost requirements rather than product preference.
Career Benefits
The credential demonstrates professional Google Cloud data-platform judgment for data engineering, analytics engineering, data architecture, and platform leadership roles. Renewal is available through the standard exam, an eligible shorter renewal exam, or designated Google Skills learning options.
Sample Questions
5 sample questions with answers and explanations. The full bank has 1,063 questions, enough for 21 full-length practice exams.
Preview — answers shown1. Your application stores user preferences (small JSON documents, <1KB each). Updates are frequent and reads must be strongly consistent. The application serves 50,000 users. What database should you use?
AMemorystore
BFirestore in native mode
CCloud SQL
DBigtable
Explanation
Firestore native mode provides strong consistency, document-oriented storage for JSON, and scales automatically. It's optimized for this use case (small documents, strong consistency, automatic scaling). Bigtable requires more complex data modeling. Cloud SQL works but requires more operational management for scaling. Memorystore is for caching, not durable primary storage.
2. A BigQuery table receives data from multiple ETL pipelines. To identify data quality issues by source, what metadata should be captured during loading?
ANo metadata needed
BAdd columns for source_system, load_timestamp, and pipeline_version during ETL processing
CUse table descriptions only
DLog to separate audit tables
Explanation
Capturing source system, load timestamp, and pipeline version as columns in the data enables filtering, aggregating, and analyzing data quality by source. Table descriptions don't allow row-level querying. Separate audit tables complicate analysis. Metadata is essential for traceability.
3. You are storing transaction records in BigQuery. Compliance requires that no records can be modified or deleted for 7 years. What should you configure?
AUse BigQuery time travel with extended retention
BExport data to Cloud Storage with Object Lock
CSet table expiration to 7 years
DEnable BigQuery table snapshots for backup
Explanation
Cloud Storage with Object Lock provides WORM (Write Once, Read Many) immutability, meeting compliance requirements. BigQuery can query the data via external tables. Table snapshots are for point-in-time copies, not immutability. Time travel provides version history but doesn't prevent deletion. Table expiration deletes data after 7 years, opposite of the requirement.
4. A Dataflow pipeline writes to BigQuery using streaming inserts. The pipeline must retry failed inserts but avoid duplicate insertions. What configuration ensures this?
AUse Dataflow's insertId generation for BigQuery streaming insert deduplication
BDisable retries
CImplement manual deduplication logic
DUse WRITE_TRUNCATE mode
Explanation
Dataflow can generate deterministic insertIds for BigQuery streaming, enabling BigQuery's built-in deduplication within the deduplication window. Retries with the same insertId don't create duplicates. Manual logic duplicates built-in functionality. Disabling retries loses data. WRITE_TRUNCATE is for batch, not streaming.
5. Solution: A company wants to mask PII data in BigQuery before sharing with analysts. They configure a Data Catalog policy tag with SHA256 hash masking on the email column. Analysts with fine-grained reader access can see the hashed emails. Does this solution meet the goal of protecting PII while allowing analysis?
AYes
BNo
Explanation
SHA256 hash masking protects the actual email values while preserving analytical utility. Analysts can group, count, and join on the hashed values without seeing actual emails. The hash is deterministic, so the same email always produces the same hash, enabling consistent analysis. This is a common privacy-preserving technique that balances protection with analytical needs. The solution correctly uses policy tags with masking rules to protect PII for users without Fine-Grained Reader access to the actual values.
Frequently Asked Questions
How many questions are on the Google Professional Data Engineer exam?
The standard exam has 40-50 multiple-choice and multiple-select questions in two hours.
What score passes Google Professional Data Engineer?
Google does not publicly disclose a passing score.
What are the current Data Engineer section weights?
The v4.2 guide lists design at about 22%, ingest and process 25%, storage 20%, prepare and use data 15%, and maintain and automate 18%.
How much does the standard exam cost?
Google lists $200 USD plus tax where applicable.
Does Professional Data Engineer have prerequisites?
No. Google recommends at least three years of industry experience, including one year designing and managing Google Cloud solutions.
How can the certification be renewed?
Eligible holders may take the standard exam, a one-hour $100 renewal exam, or complete designated Google Skills courses or skill badges.
More Google Cloud Practice Exams
Google Cloud Certified - Professional Cloud DevOps Engineer (PCDOps)
PCDOps · 1132 questions
Google Cloud Certified - Professional Cloud Network Engineer (PCNE)
PCNE · 881 questions
Google Cloud Certified - Professional Cloud Security Engineer (PCSE)
PCSE · 1075 questions
Google Cloud Certified - Professional Machine Learning Engineer (PMLE)
PMLE · 1100 questions
Google Cloud Certified - Professional Security Operations Engineer (PSOE)
PSOE · 1089 questions
Google Cloud Certified - Professional Google Workspace Administrator
PGWA · 390 questions
$17.99
One-time access to this exam
1,063 questions (21 practice exams' worth)
Unlimited timed exam simulations
Yours forever — no subscription
Detailed explanations
Free preview stays available