What are the 5 PDE exam sections and their weights?

The 5 sections are: Designing Data Processing Systems (~22%), Ingesting and Processing the Data (~25%), Storing the Data (~20%), Preparing and Using Data for Analysis (~15%), and Maintaining and Automating Data Workloads (~18%).

How is the PDE exam different from the PDE renewal exam?

The full PDE exam is 2 hours with ~60 questions including case studies (Flowlogistic, MJTelco). The renewal exam is 1 hour with 20 scenario-based questions, no case studies, and heavier emphasis on modern services like BigLake, AlloyDB, and Dataplex.

What GCP services are most heavily tested on the PDE exam?

BigQuery is the most heavily tested service across all sections. Other critical services include Dataflow (Apache Beam), Dataproc (Spark/Hadoop), Cloud Composer (Airflow), Pub/Sub, Cloud Storage, and Dataplex. The renewal exam additionally emphasizes BigLake, AlloyDB, and Dataform.

How long is the PDE exam and how many questions does it have?

The full PDE exam is 2 hours with 50-60 multiple-choice and multiple-select questions. The passing score is approximately 70-80%. The exam costs $200 USD.

Home > PDE > Exam Guide

PDE Exam Guide (2026)

The Google Cloud Professional Data Engineer exam has 5 sections: Designing Data Processing Systems (~22%), Ingesting and Processing (~25%), Storing the Data (~20%), Preparing and Using Data for Analysis (~15%), and Maintaining and Automating Data Workloads (~18%). The exam is 2 hours with 50–60 questions. BigQuery is tested in every section.

Quick Facts

Duration: 2 hours (full) / 1 hour (renewal)
Questions: 50–60 (full) / 20 (renewal)
Cost: $200 USD
Passing score: ~70–80% (Google does not publish the exact threshold)
Case studies: Flowlogistic & MJTelco (full exam only — not on renewal)

The 5 Exam Sections

Section 1: Designing Data Processing Systems (~22%)

Security, compliance, reliability, flexibility, and data migrations. You'll need to know IAM, Cloud KMS (CMEK, HSM, EKM), VPC Service Controls, disaster recovery, and migration strategies.

1.1 Designing for security and compliance (IAM, encryption, PII, data sovereignty)
1.2 Designing for reliability and fidelity (data validation, DR, fault tolerance)
1.3 Designing for flexibility and portability (multi-cloud, data governance)
1.4 Designing data migrations (BigQuery Data Transfer Service, Database Migration Service, Transfer Appliance)

Section 2: Ingesting and Processing the Data (~25%)

The heaviest section. Pipeline planning, building, and operationalizing. Expect Dataflow (Apache Beam), Dataproc, Cloud Data Fusion, Pub/Sub, and windowing/late-arriving data scenarios.

2.1 Planning the data pipelines (sources, sinks, transformation logic, encryption)
2.2 Building the pipelines (Dataflow, Dataproc, Spark, Kafka, batch vs streaming, AI enrichment)
2.3 Deploying and operationalizing (Cloud Composer, Workflows, CI/CD)

Section 3: Storing the Data (~20%)

Storage selection, data warehousing, data lakes, and data platforms. Know when to use BigQuery vs BigLake vs Bigtable vs Spanner vs Cloud SQL vs AlloyDB.

3.1 Selecting storage systems (access patterns, managed services, cost, lifecycle)
3.2 Planning for a data warehouse (data model, normalization, access patterns)
3.3 Using a data lake (discovery, processing, monitoring)
3.4 Designing for a data platform (Dataplex, Dataplex Catalog, federated governance)

Section 4: Preparing and Using Data for Analysis (~15%)

Visualization, AI/ML preparation, and data sharing. BigQuery ML, BI Engine, materialized views, Analytics Hub, and Cloud DLP masking.

4.1 Preparing data for visualization (BI Engine, materialized views, troubleshooting queries)
4.2 Preparing data for AI and ML (BigQuery ML, embeddings, RAG)
4.3 Sharing data (Analytics Hub, authorized datasets, reports)

Section 5: Maintaining and Automating Data Workloads (~18%)

Resource optimization, automation, monitoring, and fault tolerance. Cloud Monitoring, BigQuery reservations, Dataproc cluster management, and failover strategies.

5.1 Optimizing resources (cost minimization, persistent vs job-based clusters)
5.2 Designing automation and repeatability (Cloud Composer DAGs, scheduling)
5.3 Organizing workloads (BigQuery Editions, reservations, batch vs interactive)
5.4 Monitoring and troubleshooting (observability, billing issues, quotas)
5.5 Maintaining awareness of failures (fault tolerance, multi-region, data replication)

Key GCP Services to Know

Service	Role	Sections
BigQuery	Serverless data warehouse	All
Dataflow	Unified stream + batch ETL (Apache Beam)	§2, §5
Dataproc	Managed Spark/Hadoop	§2, §5
Cloud Composer	Pipeline orchestration (Airflow)	§2, §5
Pub/Sub	Event messaging	§2
BigLake	Unified analytics on multi-cloud data	§3
Dataplex	Data lake/mesh governance	§3
Cloud KMS	Key management (CMEK)	§1
Cloud DLP	PII detection & masking	§1, §4
Analytics Hub	Zero-copy data sharing	§4

How to Prepare

The PDE rewards hands-on experience with GCP's data services. Build real pipelines, query real datasets in BigQuery, and orchestrate real workflows with Cloud Composer before sitting the exam.

Review the official GCP documentation for BigQuery, Dataflow, and Dataplex.
Follow our 8-Week Study Plan for first-time candidates.
Renewing? Use the 2-Week Refresher instead.
Practice with questions verified against current Google Cloud documentation.

Start the Practice Exam