S
Google Professional Data Engineer (PDE) ACTUAL
PRACTICE EXAM 2026/2027 | Verified Questions
and Answers | Complex Architectural Scenarios &
Trade-off Analysis | Exam Pass Target - A+ Graded
SECTION 1: DESIGNING DATA PROCESSING SYSTEMS (15 Questions)
Q1: A retail company needs to store 500 TB of historical sales data for ad-hoc SQL analytics by
business analysts. Query patterns are unpredictable, with 90% of queries scanning data from the
last 12 months. Data older than 7 years must be retained for compliance but accessed rarely. Cost
optimization is critical. Which storage and lifecycle strategy should be implemented?
A. Store all data in Cloud Storage Nearline with BigQuery external tables; query directly from
Cloud Storage
B. Use BigQuery with partitioned tables by date, implement time-based partitioning with long-
term storage pricing for partitions older than 90 days, and configure table expiration for data
older than 7 years [CORRECT]
C. Store data in Cloud Bigtable with column families for different time ranges; use Dataflow to
move cold data to Cloud Storage Coldline
D. Use Cloud SQL PostgreSQL with read replicas; implement manual data archival to Cloud
Storage using cron jobs
Correct Answer: B
Q2: A financial services firm needs to process real-time stock market data (1 million
events/second) with sub-100ms latency for trading decisions. Data must be queryable by stock
symbol and time range with high throughput. Historical data must be retained for 10 years.
Which architecture meets these requirements?
A. Pub/Sub → Dataflow → BigQuery with streaming inserts
B. Pub/Sub → Dataflow → Cloud Bigtable for hot data; export to Cloud Storage for archival
[CORRECT]
C. Cloud SQL with high-memory instances; use read replicas for query scaling
,S
D. Firestore in Native mode with composite indexes on symbol and timestamp
Correct Answer: B
Q3: A healthcare organization needs to store patient records with strict HIPAA compliance
requirements. Data must be encrypted at rest with customer-managed keys, and access must be
audited. Analytics queries will run monthly on de-identified data. Which solution provides the
required security controls?
A. BigQuery with default encryption and IAM policies
B. BigQuery with CMEK (Customer-Managed Encryption Keys), VPC Service Controls, and
Cloud Audit Logs with Data Access logs enabled [CORRECT]
C. Cloud Storage with standard encryption and bucket-level IAM
D. Dataproc with local SSD encryption and Kerberos authentication
Correct Answer: B
Q4: A gaming company needs to analyze player behavior patterns from JSON event logs. Data
volume is 50 TB/day with nested and repeated fields. Analysts need to run complex aggregations
and joins across multiple event types. Which storage solution is most appropriate?
A. Cloud SQL with JSON data type and indexes
B. BigQuery with nested/repeated fields and standard SQL support for semi-structured data
[CORRECT]
C. Cloud Bigtable with column qualifiers for nested fields
D. Firestore with collection group queries
Correct Answer: B
Q5: An e-commerce platform needs to migrate a 20 TB PostgreSQL database to GCP with
minimal downtime. The database has complex stored procedures and requires ACID transactions.
Post-migration, the database must support 50,000 QPS with global read replicas. Which
migration path should be chosen?
A. Export to CSV, load into BigQuery, rewrite queries in standard SQL
B. Use Database Migration Service to migrate to Cloud SQL for PostgreSQL, then add read
replicas; later evaluate Cloud Spanner if horizontal write scaling is needed [CORRECT]
, S
C. Migrate to Cloud Bigtable using Dataflow for schema transformation
D. Replatform to Firestore with Datastore mode for automatic sharding
Correct Answer: B
Q6: A logistics company needs to store GPS tracking data from 100,000 vehicles, updating
location every 10 seconds. Queries need to retrieve the latest position for any vehicle and
calculate travel routes for specific time windows. Which database combination is optimal?
A. BigQuery for all storage and queries; use streaming inserts for real-time data
B. Cloud Bigtable for real-time writes and recent data queries; BigQuery for historical analytics
via Dataflow export [CORRECT]
C. Cloud SQL with partitioning by vehicle_id; use read replicas for query performance
D. Firestore with TTL policies for old data; use collection group queries for analytics
Correct Answer: B
Q7: A media company needs to process 10 TB of video content daily for transcoding and
metadata extraction. Processing is batch-oriented with no strict latency requirements. Cost per
GB processed must be minimized. Which architecture is most cost-effective?
A. Pub/Sub → Dataflow with autoscaling → Cloud Storage
B. Cloud Storage → Dataproc with preemptible VMs → Cloud Storage [CORRECT]
C. Cloud Storage → Cloud Run with concurrent processing → Cloud Storage
D. Cloud Storage → BigQuery with user-defined functions → Cloud Storage
Correct Answer: B
Q8: A manufacturing company needs to design a data lake for IoT sensor data. Data arrives in
multiple formats (JSON, Avro, Parquet) from different systems. Data scientists need to query raw
data, while analysts need cleaned data. Which architecture supports both requirements
efficiently?
A. Store all data in BigQuery native tables; create views for data scientists
B. Use Cloud Storage as data lake with raw and processed zones; BigQuery external tables for
raw data access, native tables for cleaned data via Dataflow [CORRECT]
Google Professional Data Engineer (PDE) ACTUAL
PRACTICE EXAM 2026/2027 | Verified Questions
and Answers | Complex Architectural Scenarios &
Trade-off Analysis | Exam Pass Target - A+ Graded
SECTION 1: DESIGNING DATA PROCESSING SYSTEMS (15 Questions)
Q1: A retail company needs to store 500 TB of historical sales data for ad-hoc SQL analytics by
business analysts. Query patterns are unpredictable, with 90% of queries scanning data from the
last 12 months. Data older than 7 years must be retained for compliance but accessed rarely. Cost
optimization is critical. Which storage and lifecycle strategy should be implemented?
A. Store all data in Cloud Storage Nearline with BigQuery external tables; query directly from
Cloud Storage
B. Use BigQuery with partitioned tables by date, implement time-based partitioning with long-
term storage pricing for partitions older than 90 days, and configure table expiration for data
older than 7 years [CORRECT]
C. Store data in Cloud Bigtable with column families for different time ranges; use Dataflow to
move cold data to Cloud Storage Coldline
D. Use Cloud SQL PostgreSQL with read replicas; implement manual data archival to Cloud
Storage using cron jobs
Correct Answer: B
Q2: A financial services firm needs to process real-time stock market data (1 million
events/second) with sub-100ms latency for trading decisions. Data must be queryable by stock
symbol and time range with high throughput. Historical data must be retained for 10 years.
Which architecture meets these requirements?
A. Pub/Sub → Dataflow → BigQuery with streaming inserts
B. Pub/Sub → Dataflow → Cloud Bigtable for hot data; export to Cloud Storage for archival
[CORRECT]
C. Cloud SQL with high-memory instances; use read replicas for query scaling
,S
D. Firestore in Native mode with composite indexes on symbol and timestamp
Correct Answer: B
Q3: A healthcare organization needs to store patient records with strict HIPAA compliance
requirements. Data must be encrypted at rest with customer-managed keys, and access must be
audited. Analytics queries will run monthly on de-identified data. Which solution provides the
required security controls?
A. BigQuery with default encryption and IAM policies
B. BigQuery with CMEK (Customer-Managed Encryption Keys), VPC Service Controls, and
Cloud Audit Logs with Data Access logs enabled [CORRECT]
C. Cloud Storage with standard encryption and bucket-level IAM
D. Dataproc with local SSD encryption and Kerberos authentication
Correct Answer: B
Q4: A gaming company needs to analyze player behavior patterns from JSON event logs. Data
volume is 50 TB/day with nested and repeated fields. Analysts need to run complex aggregations
and joins across multiple event types. Which storage solution is most appropriate?
A. Cloud SQL with JSON data type and indexes
B. BigQuery with nested/repeated fields and standard SQL support for semi-structured data
[CORRECT]
C. Cloud Bigtable with column qualifiers for nested fields
D. Firestore with collection group queries
Correct Answer: B
Q5: An e-commerce platform needs to migrate a 20 TB PostgreSQL database to GCP with
minimal downtime. The database has complex stored procedures and requires ACID transactions.
Post-migration, the database must support 50,000 QPS with global read replicas. Which
migration path should be chosen?
A. Export to CSV, load into BigQuery, rewrite queries in standard SQL
B. Use Database Migration Service to migrate to Cloud SQL for PostgreSQL, then add read
replicas; later evaluate Cloud Spanner if horizontal write scaling is needed [CORRECT]
, S
C. Migrate to Cloud Bigtable using Dataflow for schema transformation
D. Replatform to Firestore with Datastore mode for automatic sharding
Correct Answer: B
Q6: A logistics company needs to store GPS tracking data from 100,000 vehicles, updating
location every 10 seconds. Queries need to retrieve the latest position for any vehicle and
calculate travel routes for specific time windows. Which database combination is optimal?
A. BigQuery for all storage and queries; use streaming inserts for real-time data
B. Cloud Bigtable for real-time writes and recent data queries; BigQuery for historical analytics
via Dataflow export [CORRECT]
C. Cloud SQL with partitioning by vehicle_id; use read replicas for query performance
D. Firestore with TTL policies for old data; use collection group queries for analytics
Correct Answer: B
Q7: A media company needs to process 10 TB of video content daily for transcoding and
metadata extraction. Processing is batch-oriented with no strict latency requirements. Cost per
GB processed must be minimized. Which architecture is most cost-effective?
A. Pub/Sub → Dataflow with autoscaling → Cloud Storage
B. Cloud Storage → Dataproc with preemptible VMs → Cloud Storage [CORRECT]
C. Cloud Storage → Cloud Run with concurrent processing → Cloud Storage
D. Cloud Storage → BigQuery with user-defined functions → Cloud Storage
Correct Answer: B
Q8: A manufacturing company needs to design a data lake for IoT sensor data. Data arrives in
multiple formats (JSON, Avro, Parquet) from different systems. Data scientists need to query raw
data, while analysts need cleaned data. Which architecture supports both requirements
efficiently?
A. Store all data in BigQuery native tables; create views for data scientists
B. Use Cloud Storage as data lake with raw and processed zones; BigQuery external tables for
raw data access, native tables for cleaned data via Dataflow [CORRECT]
