GCP PROFESSIONAL DATA ENGINEER CERTIFICATION | HIGHER EDUCATION CLOUD COMPUTING
2026/2027 ACADEMIC YEAR | VERIFIED Q&A | ADVANCED DATA ENGINEERING & CLOUD ARCHITECTURE
GCP DATA
ENGINEER
PROFESSIONAL 2026/2027
GCP Professional Data Engineer Certification
Higher Education Cloud Computing Curriculum
2026/2027 Academic Year | Verified Q&A for Advanced Data Engineering
and Cloud Architecture Learners | Actual Verified Exam
150Q VERIFIED NEWEST EXAM RATIONALES
ACTUAL EXAM 2026/2027 ADV LEARNERS
INCLUDES:
• BigQuery, Dataflow, Dataproc, Pub/Sub, Data Fusion, Composer, Storage
• Data Modeling, ETL/ELT, ML Pipelines, Security, Governance, Performance
• 150 Verified Questions + Answers + Detailed Rationales | Professional Data Engineer
• Professional Study Guide | Borders + Page Numbers | 2026/2027 Edition
GCP Professional Data Engineer | Google Cloud | Not affiliated with Google
Confidential Study Guide - Educational Purposes Only | 2026/2027 Edition
,BigQuery Dataflow Pub/Sub Dataproc Cloud Storage
Description: BigQuery serverless data warehouse architecture separation storage compute, Dremel columnar storage, Borg slots, architecture projects datasets tables
views, optimization best practices partitioning ingestion-time column range, clustering up to 4 columns, slot allocation baseline autoscaling, reservations, cost control avoid
SELECT * use SELECT specific columns, use approximate aggregation APPROX_COUNT_DISTINCT, materialized views, BI Engine, Dataflow managed Apache Beam
runner unified batch streaming programming model PCollection PTransform Pipeline, windowing fixed sliding session, triggers early late, watermarks, streaming
exactly-once processing with checkpointing Pub/Sub to BigQuery pipeline, auto scaling horizontal, right fitting, Pub/Sub async messaging at-least-once default
exactly-once delivery with ordering keys and enable_exactly_once_delivery, push pull subscriptions, schemas, dead letter topic, retention 7 days default max 31 days,
ordering keys, Dataproc managed Hadoop Spark Flink cluster ephemeral vs long-running, autoscaling policy, initialization actions scripts, component gateway, best
practices use preemptible secondary workers cost saving, store data Cloud Storage GCS not HDFS ephemeral, use Dataproc Serverless, Cloud Storage classes Standard
for frequent access Nearline 30 days minimum Coldline 90 days Archive 365 days, lifecycle policies age based delete transition to lower class, object versioning, retention
policies, strong consistency, dual multi-region.
Data Modeling ETL/ELT Data Fusion Composer Orchestration
Description: Data modeling star schema fact table quantitative metrics sales amount, dimension tables descriptive attributes product customer date, snowflake normalized
dimensions, BigQuery denormalization best practice nested repeated fields STRUCT ARRAY due to columnar storage avoids joins, data warehouse modeling, Data Vault,
ETL Extract Transform Load traditional Data Fusion managed CDAP Cloud Data Fusion visual ETL Wrangler pipelines, ELT Extract Load Transform modern BigQuery
transformation dbt Dataform, ingestion batch streaming Data Transfer Service, Composer managed Airflow orchestration DAG Directed Acyclic Graph tasks operators
sensors, Composer 2 autoscaling, best practices idempotent DAGs, XComs, monitoring, Storage data lake architecture bronze silver gold zones raw cleaned curated, file
formats Avro Parquet ORC columnar, partitioning strategies.
ML Pipelines Vertex AI Security Governance Performance
Description: ML pipelines Vertex AI platform custom training AutoML Tabular Vision NLP, BigQuery ML SQL-based machine learning CREATE MODEL model types linear
regression logistic regression k-means DNN XGBoost, feature store Vertex AI Feature Store offline online serving, Kubeflow Pipelines TFX TensorFlow Extended, training
serving skew, security governance IAM Identity Access Management roles primitive Owner Editor Viewer predefined roles Data Engineer BigQuery Admin custom roles
least privilege principle, service accounts impersonation, VPC Service Controls perimeter protects data exfiltration defines security perimeter projects buckets, Private
Google Access Private Service Connect, DLP Data Loss Prevention API discover classify de-identify PII infoTypes inspection templates, CMEK Customer Managed
Encryption Keys Cloud KMS, audit logs, performance troubleshooting data skew keys uneven distribution causes stragglers salting technique adding random prefix
re-aggregating, hotspots BigQuery slot contention shuffle stage, monitoring Cloud Monitoring Cloud Logging Dataflow job metrics system lag data freshness, optimization
query plan EXPLAIN, cost optimization committed use discounts, reliability design.
Page 2 - GCP Data Engineer 150Q 2026/2027
,Question 1: Q1: BigQuery - architecture and optimization best practices?
A. BigQuery serverless data warehouse columnar storage separation storage compute, optimization partitioning clustering, slot allocation, best practices
avoid SELECT *, use approximate aggregation
B. BigQuery row storage only
C. No optimization BigQuery
D. Only SELECT * BigQuery
CORRECT ANSWER: A. BigQuery serverless data warehouse columnar storage separation storage compute, optimization partitioning
clustering, slot allocation, best practices avoid SELECT *, use approximate aggregation
RATIONALE:
Rationale: BigQuery serverless data warehouse columnar storage separation storage compute, optimization partitioning clustering, slot allocation, best practices avoid
SELECT *, use approximate aggregation. Per Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam,
Pub/Sub, Dataproc, Cloud Storage, data modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP,
performance troubleshooting, for advanced data engineering and cloud architecture learners.
Question 2: Q2: Dataflow - Apache Beam and streaming vs batch?
A. Dataflow only streaming
B. Dataflow only batch
C. Dataflow managed Apache Beam runner, unified batch streaming, windowing triggers watermarks, auto scaling, streaming exactly-once processing
Pub/Sub to BigQuery
D. No Beam Dataflow
CORRECT ANSWER: C. Dataflow managed Apache Beam runner, unified batch streaming, windowing triggers watermarks, auto scaling,
streaming exactly-once processing Pub/Sub to BigQuery
RATIONALE:
Rationale: Dataflow managed Apache Beam runner, unified batch streaming, windowing triggers watermarks, auto scaling, streaming exactly-once processing Pub/Sub to
BigQuery. Per Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud
Storage, data modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for
advanced data engineering and cloud architecture learners.
Question 3: Q3: Pub/Sub - messaging and ordering exactly-once?
A. No ordering Pub/Sub
B. Pub/Sub only sync messaging
C. No exactly-once Pub/Sub
D. Pub/Sub async messaging at-least-once default, exactly-once with ordering keys enabled, push pull subscriptions, dead letter topic, retention 7 days
default 31 max
CORRECT ANSWER: D. Pub/Sub async messaging at-least-once default, exactly-once with ordering keys enabled, push pull subscriptions,
dead letter topic, retention 7 days default 31 max
RATIONALE:
Rationale: Pub/Sub async messaging at-least-once default, exactly-once with ordering keys enabled, push pull subscriptions, dead letter topic, retention 7 days default 31
max. Per Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud Storage,
data modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for
advanced data engineering and cloud architecture learners.
Question 4: Q4: Dataproc - managed Hadoop Spark and best practices?
A. Dataproc only Hadoop no Spark
B. Dataproc managed Hadoop Spark cluster, ephemeral clusters, autoscaling, initialization actions, best practices use preemptible workers, store data
GCS not HDFS
C. No autoscaling Dataproc
D. Only permanent clusters Dataproc
CORRECT ANSWER: B. Dataproc managed Hadoop Spark cluster, ephemeral clusters, autoscaling, initialization actions, best practices use
preemptible workers, store data GCS not HDFS
RATIONALE:
Rationale: Dataproc managed Hadoop Spark cluster, ephemeral clusters, autoscaling, initialization actions, best practices use preemptible workers, store data GCS not
HDFS. Per Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud
Storage, data modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for
advanced data engineering and cloud architecture learners.
Question 5: Q5: Cloud Storage - classes and lifecycle?
A. Cloud Storage classes Standard Nearline Coldline Archive, lifecycle policies delete transition, object versioning, strong consistency
B. Only Standard Storage class
C. No lifecycle Storage
D. No classes Cloud Storage
CORRECT ANSWER: A. Cloud Storage classes Standard Nearline Coldline Archive, lifecycle policies delete transition, object versioning,
strong consistency
RATIONALE:
Rationale: Cloud Storage classes Standard Nearline Coldline Archive, lifecycle policies delete transition, object versioning, strong consistency. Per Google Cloud
Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud Storage, data modeling, ETL/ELT
Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for advanced data engineering and
cloud architecture learners.
Question 6: Q6: Data modeling - star schema and BigQuery denormalization?
A. No star schema Data modeling
B. Only normalized model BigQuery
Page 3 - GCP Data Engineer 150Q 2026/2027
, C. Data modeling star schema fact dimension, BigQuery denormalization nested repeated fields best due to columnar, avoid excessive joins, use
STRUCT ARRAY
D. Only 3NF BigQuery
CORRECT ANSWER: C. Data modeling star schema fact dimension, BigQuery denormalization nested repeated fields best due to columnar,
avoid excessive joins, use STRUCT ARRAY
RATIONALE:
Rationale: Data modeling star schema fact dimension, BigQuery denormalization nested repeated fields best due to columnar, avoid excessive joins, use STRUCT
ARRAY. Per Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud
Storage, data modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for
advanced data engineering and cloud architecture learners.
Question 7: Q7: ETL/ELT - Data Fusion and Composer orchestration?
A. No Data Fusion ETL
B. Only ETL no ELT
C. No Composer orchestration
D. ETL Extract Transform Load Data Fusion managed CDAP, ELT Extract Load Transform BigQuery transformation, Composer managed Airflow DAG
orchestration
CORRECT ANSWER: D. ETL Extract Transform Load Data Fusion managed CDAP, ELT Extract Load Transform BigQuery transformation,
Composer managed Airflow DAG orchestration
RATIONALE:
Rationale: ETL Extract Transform Load Data Fusion managed CDAP, ELT Extract Load Transform BigQuery transformation, Composer managed Airflow DAG
orchestration. Per Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud
Storage, data modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for
advanced data engineering and cloud architecture learners.
Question 8: Q8: ML pipelines - Vertex AI and BigQuery ML?
A. No ML pipelines GCP
B. ML pipelines Vertex AI custom training AutoML, BigQuery ML SQL-based ML CREATE MODEL, feature store, Kubeflow Pipelines, TFX
C. Only BigQuery ML no Vertex AI
D. No Vertex AI GCP
CORRECT ANSWER: B. ML pipelines Vertex AI custom training AutoML, BigQuery ML SQL-based ML CREATE MODEL, feature store,
Kubeflow Pipelines, TFX
RATIONALE:
Rationale: ML pipelines Vertex AI custom training AutoML, BigQuery ML SQL-based ML CREATE MODEL, feature store, Kubeflow Pipelines, TFX. Per Google Cloud
Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud Storage, data modeling, ETL/ELT
Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for advanced data engineering and
cloud architecture learners.
Question 9: Q9: Security governance - IAM VPC Service Controls DLP?
A. Security IAM roles primitive predefined custom least privilege, VPC Service Controls perimeter, DLP Data Loss Prevention API PII de-identification,
CMEK
B. No IAM security GCP
C. No VPC SC security
D. No DLP security GCP
CORRECT ANSWER: A. Security IAM roles primitive predefined custom least privilege, VPC Service Controls perimeter, DLP Data Loss
Prevention API PII de-identification, CMEK
RATIONALE:
Rationale: Security IAM roles primitive predefined custom least privilege, VPC Service Controls perimeter, DLP Data Loss Prevention API PII de-identification, CMEK. Per
Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud Storage, data
modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for advanced
data engineering and cloud architecture learners.
Question 10: Q10: Performance troubleshooting - data skew and hotspots?
A. No hotspots performance
B. No data skew performance
C. Performance troubleshooting data skew keys uneven distribution salting, hotspots BigQuery slot contention shuffle, monitoring Cloud Monitoring
Dataflow job metrics
D. No troubleshooting performance
CORRECT ANSWER: C. Performance troubleshooting data skew keys uneven distribution salting, hotspots BigQuery slot contention shuffle,
monitoring Cloud Monitoring Dataflow job metrics
RATIONALE:
Rationale: Performance troubleshooting data skew keys uneven distribution salting, hotspots BigQuery slot contention shuffle, monitoring Cloud Monitoring Dataflow job
metrics. Per Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud
Storage, data modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for
advanced data engineering and cloud architecture learners.
Page 4 - GCP Data Engineer 150Q 2026/2027
2026/2027 ACADEMIC YEAR | VERIFIED Q&A | ADVANCED DATA ENGINEERING & CLOUD ARCHITECTURE
GCP DATA
ENGINEER
PROFESSIONAL 2026/2027
GCP Professional Data Engineer Certification
Higher Education Cloud Computing Curriculum
2026/2027 Academic Year | Verified Q&A for Advanced Data Engineering
and Cloud Architecture Learners | Actual Verified Exam
150Q VERIFIED NEWEST EXAM RATIONALES
ACTUAL EXAM 2026/2027 ADV LEARNERS
INCLUDES:
• BigQuery, Dataflow, Dataproc, Pub/Sub, Data Fusion, Composer, Storage
• Data Modeling, ETL/ELT, ML Pipelines, Security, Governance, Performance
• 150 Verified Questions + Answers + Detailed Rationales | Professional Data Engineer
• Professional Study Guide | Borders + Page Numbers | 2026/2027 Edition
GCP Professional Data Engineer | Google Cloud | Not affiliated with Google
Confidential Study Guide - Educational Purposes Only | 2026/2027 Edition
,BigQuery Dataflow Pub/Sub Dataproc Cloud Storage
Description: BigQuery serverless data warehouse architecture separation storage compute, Dremel columnar storage, Borg slots, architecture projects datasets tables
views, optimization best practices partitioning ingestion-time column range, clustering up to 4 columns, slot allocation baseline autoscaling, reservations, cost control avoid
SELECT * use SELECT specific columns, use approximate aggregation APPROX_COUNT_DISTINCT, materialized views, BI Engine, Dataflow managed Apache Beam
runner unified batch streaming programming model PCollection PTransform Pipeline, windowing fixed sliding session, triggers early late, watermarks, streaming
exactly-once processing with checkpointing Pub/Sub to BigQuery pipeline, auto scaling horizontal, right fitting, Pub/Sub async messaging at-least-once default
exactly-once delivery with ordering keys and enable_exactly_once_delivery, push pull subscriptions, schemas, dead letter topic, retention 7 days default max 31 days,
ordering keys, Dataproc managed Hadoop Spark Flink cluster ephemeral vs long-running, autoscaling policy, initialization actions scripts, component gateway, best
practices use preemptible secondary workers cost saving, store data Cloud Storage GCS not HDFS ephemeral, use Dataproc Serverless, Cloud Storage classes Standard
for frequent access Nearline 30 days minimum Coldline 90 days Archive 365 days, lifecycle policies age based delete transition to lower class, object versioning, retention
policies, strong consistency, dual multi-region.
Data Modeling ETL/ELT Data Fusion Composer Orchestration
Description: Data modeling star schema fact table quantitative metrics sales amount, dimension tables descriptive attributes product customer date, snowflake normalized
dimensions, BigQuery denormalization best practice nested repeated fields STRUCT ARRAY due to columnar storage avoids joins, data warehouse modeling, Data Vault,
ETL Extract Transform Load traditional Data Fusion managed CDAP Cloud Data Fusion visual ETL Wrangler pipelines, ELT Extract Load Transform modern BigQuery
transformation dbt Dataform, ingestion batch streaming Data Transfer Service, Composer managed Airflow orchestration DAG Directed Acyclic Graph tasks operators
sensors, Composer 2 autoscaling, best practices idempotent DAGs, XComs, monitoring, Storage data lake architecture bronze silver gold zones raw cleaned curated, file
formats Avro Parquet ORC columnar, partitioning strategies.
ML Pipelines Vertex AI Security Governance Performance
Description: ML pipelines Vertex AI platform custom training AutoML Tabular Vision NLP, BigQuery ML SQL-based machine learning CREATE MODEL model types linear
regression logistic regression k-means DNN XGBoost, feature store Vertex AI Feature Store offline online serving, Kubeflow Pipelines TFX TensorFlow Extended, training
serving skew, security governance IAM Identity Access Management roles primitive Owner Editor Viewer predefined roles Data Engineer BigQuery Admin custom roles
least privilege principle, service accounts impersonation, VPC Service Controls perimeter protects data exfiltration defines security perimeter projects buckets, Private
Google Access Private Service Connect, DLP Data Loss Prevention API discover classify de-identify PII infoTypes inspection templates, CMEK Customer Managed
Encryption Keys Cloud KMS, audit logs, performance troubleshooting data skew keys uneven distribution causes stragglers salting technique adding random prefix
re-aggregating, hotspots BigQuery slot contention shuffle stage, monitoring Cloud Monitoring Cloud Logging Dataflow job metrics system lag data freshness, optimization
query plan EXPLAIN, cost optimization committed use discounts, reliability design.
Page 2 - GCP Data Engineer 150Q 2026/2027
,Question 1: Q1: BigQuery - architecture and optimization best practices?
A. BigQuery serverless data warehouse columnar storage separation storage compute, optimization partitioning clustering, slot allocation, best practices
avoid SELECT *, use approximate aggregation
B. BigQuery row storage only
C. No optimization BigQuery
D. Only SELECT * BigQuery
CORRECT ANSWER: A. BigQuery serverless data warehouse columnar storage separation storage compute, optimization partitioning
clustering, slot allocation, best practices avoid SELECT *, use approximate aggregation
RATIONALE:
Rationale: BigQuery serverless data warehouse columnar storage separation storage compute, optimization partitioning clustering, slot allocation, best practices avoid
SELECT *, use approximate aggregation. Per Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam,
Pub/Sub, Dataproc, Cloud Storage, data modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP,
performance troubleshooting, for advanced data engineering and cloud architecture learners.
Question 2: Q2: Dataflow - Apache Beam and streaming vs batch?
A. Dataflow only streaming
B. Dataflow only batch
C. Dataflow managed Apache Beam runner, unified batch streaming, windowing triggers watermarks, auto scaling, streaming exactly-once processing
Pub/Sub to BigQuery
D. No Beam Dataflow
CORRECT ANSWER: C. Dataflow managed Apache Beam runner, unified batch streaming, windowing triggers watermarks, auto scaling,
streaming exactly-once processing Pub/Sub to BigQuery
RATIONALE:
Rationale: Dataflow managed Apache Beam runner, unified batch streaming, windowing triggers watermarks, auto scaling, streaming exactly-once processing Pub/Sub to
BigQuery. Per Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud
Storage, data modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for
advanced data engineering and cloud architecture learners.
Question 3: Q3: Pub/Sub - messaging and ordering exactly-once?
A. No ordering Pub/Sub
B. Pub/Sub only sync messaging
C. No exactly-once Pub/Sub
D. Pub/Sub async messaging at-least-once default, exactly-once with ordering keys enabled, push pull subscriptions, dead letter topic, retention 7 days
default 31 max
CORRECT ANSWER: D. Pub/Sub async messaging at-least-once default, exactly-once with ordering keys enabled, push pull subscriptions,
dead letter topic, retention 7 days default 31 max
RATIONALE:
Rationale: Pub/Sub async messaging at-least-once default, exactly-once with ordering keys enabled, push pull subscriptions, dead letter topic, retention 7 days default 31
max. Per Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud Storage,
data modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for
advanced data engineering and cloud architecture learners.
Question 4: Q4: Dataproc - managed Hadoop Spark and best practices?
A. Dataproc only Hadoop no Spark
B. Dataproc managed Hadoop Spark cluster, ephemeral clusters, autoscaling, initialization actions, best practices use preemptible workers, store data
GCS not HDFS
C. No autoscaling Dataproc
D. Only permanent clusters Dataproc
CORRECT ANSWER: B. Dataproc managed Hadoop Spark cluster, ephemeral clusters, autoscaling, initialization actions, best practices use
preemptible workers, store data GCS not HDFS
RATIONALE:
Rationale: Dataproc managed Hadoop Spark cluster, ephemeral clusters, autoscaling, initialization actions, best practices use preemptible workers, store data GCS not
HDFS. Per Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud
Storage, data modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for
advanced data engineering and cloud architecture learners.
Question 5: Q5: Cloud Storage - classes and lifecycle?
A. Cloud Storage classes Standard Nearline Coldline Archive, lifecycle policies delete transition, object versioning, strong consistency
B. Only Standard Storage class
C. No lifecycle Storage
D. No classes Cloud Storage
CORRECT ANSWER: A. Cloud Storage classes Standard Nearline Coldline Archive, lifecycle policies delete transition, object versioning,
strong consistency
RATIONALE:
Rationale: Cloud Storage classes Standard Nearline Coldline Archive, lifecycle policies delete transition, object versioning, strong consistency. Per Google Cloud
Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud Storage, data modeling, ETL/ELT
Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for advanced data engineering and
cloud architecture learners.
Question 6: Q6: Data modeling - star schema and BigQuery denormalization?
A. No star schema Data modeling
B. Only normalized model BigQuery
Page 3 - GCP Data Engineer 150Q 2026/2027
, C. Data modeling star schema fact dimension, BigQuery denormalization nested repeated fields best due to columnar, avoid excessive joins, use
STRUCT ARRAY
D. Only 3NF BigQuery
CORRECT ANSWER: C. Data modeling star schema fact dimension, BigQuery denormalization nested repeated fields best due to columnar,
avoid excessive joins, use STRUCT ARRAY
RATIONALE:
Rationale: Data modeling star schema fact dimension, BigQuery denormalization nested repeated fields best due to columnar, avoid excessive joins, use STRUCT
ARRAY. Per Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud
Storage, data modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for
advanced data engineering and cloud architecture learners.
Question 7: Q7: ETL/ELT - Data Fusion and Composer orchestration?
A. No Data Fusion ETL
B. Only ETL no ELT
C. No Composer orchestration
D. ETL Extract Transform Load Data Fusion managed CDAP, ELT Extract Load Transform BigQuery transformation, Composer managed Airflow DAG
orchestration
CORRECT ANSWER: D. ETL Extract Transform Load Data Fusion managed CDAP, ELT Extract Load Transform BigQuery transformation,
Composer managed Airflow DAG orchestration
RATIONALE:
Rationale: ETL Extract Transform Load Data Fusion managed CDAP, ELT Extract Load Transform BigQuery transformation, Composer managed Airflow DAG
orchestration. Per Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud
Storage, data modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for
advanced data engineering and cloud architecture learners.
Question 8: Q8: ML pipelines - Vertex AI and BigQuery ML?
A. No ML pipelines GCP
B. ML pipelines Vertex AI custom training AutoML, BigQuery ML SQL-based ML CREATE MODEL, feature store, Kubeflow Pipelines, TFX
C. Only BigQuery ML no Vertex AI
D. No Vertex AI GCP
CORRECT ANSWER: B. ML pipelines Vertex AI custom training AutoML, BigQuery ML SQL-based ML CREATE MODEL, feature store,
Kubeflow Pipelines, TFX
RATIONALE:
Rationale: ML pipelines Vertex AI custom training AutoML, BigQuery ML SQL-based ML CREATE MODEL, feature store, Kubeflow Pipelines, TFX. Per Google Cloud
Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud Storage, data modeling, ETL/ELT
Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for advanced data engineering and
cloud architecture learners.
Question 9: Q9: Security governance - IAM VPC Service Controls DLP?
A. Security IAM roles primitive predefined custom least privilege, VPC Service Controls perimeter, DLP Data Loss Prevention API PII de-identification,
CMEK
B. No IAM security GCP
C. No VPC SC security
D. No DLP security GCP
CORRECT ANSWER: A. Security IAM roles primitive predefined custom least privilege, VPC Service Controls perimeter, DLP Data Loss
Prevention API PII de-identification, CMEK
RATIONALE:
Rationale: Security IAM roles primitive predefined custom least privilege, VPC Service Controls perimeter, DLP Data Loss Prevention API PII de-identification, CMEK. Per
Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud Storage, data
modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for advanced
data engineering and cloud architecture learners.
Question 10: Q10: Performance troubleshooting - data skew and hotspots?
A. No hotspots performance
B. No data skew performance
C. Performance troubleshooting data skew keys uneven distribution salting, hotspots BigQuery slot contention shuffle, monitoring Cloud Monitoring
Dataflow job metrics
D. No troubleshooting performance
CORRECT ANSWER: C. Performance troubleshooting data skew keys uneven distribution salting, hotspots BigQuery slot contention shuffle,
monitoring Cloud Monitoring Dataflow job metrics
RATIONALE:
Rationale: Performance troubleshooting data skew keys uneven distribution salting, hotspots BigQuery slot contention shuffle, monitoring Cloud Monitoring Dataflow job
metrics. Per Google Cloud Professional Data Engineer certification exam guide 2026/2027, covering BigQuery, Dataflow Apache Beam, Pub/Sub, Dataproc, Cloud
Storage, data modeling, ETL/ELT Data Fusion Composer, ML pipelines Vertex AI BigQuery ML, security governance IAM VPC-SC DLP, performance troubleshooting, for
advanced data engineering and cloud architecture learners.
Page 4 - GCP Data Engineer 150Q 2026/2027
