Courseiva

Google Professional Data Engineer (PDE) — Questions 76–150

890 questions total · 12pages · All types, answers revealed

Take a mock examExam hub
Page 1

Page 2 of 12

Page 3
76
MCQeasy

What is the primary purpose of Vertex AI Feature Store?

A.To manage and track ML experiments
B.To train machine learning models using AutoML
C.To transform raw data into features using SQL
D.To store and serve features for machine learning models at scale
AnswerD

Feature Store is designed for feature management and serving.

Why this answer

Vertex AI Feature Store is a managed service for storing, serving, and sharing ML features. It supports both online (low-latency serving for prediction) and offline (batch serving for training) serving. It is not for model training, data transformation, or experiment tracking.

Full explanation →
77
MCQhard

A company uses Dataplex to manage data quality across multiple BigQuery datasets. They want to define a data quality rule that checks if a column 'email' contains a valid email format. Which Dataplex feature should they use?

A.Use Cloud DLP to classify and validate emails.
B.Use the built-in 'email' rule type in Dataplex.
C.Create a custom Data Quality rule using the 'regex' type.
D.Create a Dataflow pipeline to validate emails and write results to a separate table.
AnswerC

Dataplex Data Quality supports regex rules to validate formats like email.

Why this answer

Dataplex Data Quality allows predefined rule types including a 'regex' rule for pattern matching. There is no built-in 'email' rule, so a regex check is appropriate.

Full explanation →
78
MCQeasy

A company needs to process real-time clickstream data and store it in a data warehouse for SQL-based analytics. The data volume is moderate. Which combination of Google Cloud services is most cost-effective?

A.Cloud Pub/Sub, Cloud Dataproc, Cloud Storage
B.Cloud Pub/Sub, Cloud Dataflow, Cloud Spanner
C.Cloud Pub/Sub, Cloud Dataflow, BigQuery
D.Cloud Pub/Sub, Cloud Dataflow, Cloud Storage
AnswerC

Best for real-time SQL analytics.

Why this answer

Cloud Pub/Sub ingests real-time clickstream data, Cloud Dataflow processes it with low latency, and BigQuery provides a serverless, SQL-based data warehouse that is cost-effective for moderate data volumes due to its pay-per-query pricing and automatic scaling. This combination avoids the overhead of managing clusters (Dataproc) or expensive storage (Cloud Spanner) while directly supporting SQL analytics.

Exam trap

Google Cloud often tests the misconception that Cloud Storage is a suitable destination for analytics-ready data, but it lacks native SQL querying, forcing candidates to overlook BigQuery's direct integration with Dataflow for real-time analytics.

How to eliminate wrong answers

Option A is wrong because Cloud Dataproc requires a running cluster (even with preemptible VMs) and is optimized for batch processing, not real-time streaming, and Cloud Storage is not a SQL-queryable data warehouse, forcing additional ETL steps. Option B is wrong because Cloud Spanner is a globally distributed, strongly consistent relational database designed for transactional workloads, not cost-effective for analytics at moderate data volumes; its per-node pricing makes it expensive compared to BigQuery's serverless model. Option D is wrong because Cloud Storage is an object store, not a data warehouse; storing processed data there would require additional services (e.g., BigQuery external tables or Dataproc) to run SQL analytics, increasing complexity and cost.

Full explanation →
79
MCQmedium

You are building a forecasting model to predict daily sales for the next 90 days using historical sales data with clear seasonality and trend. You want to use BigQuery ML with minimal manual tuning. Which model type should you choose?

A.ARIMA
B.Boosted tree (XGBoost)
C.ARIMA_PLUS
D.Linear regression
AnswerC

ARIMA_PLUS automatically detects seasonality, trend, and holiday effects; ideal for time-series forecasting without manual tuning.

Why this answer

ARIMA_PLUS is specifically designed for time-series forecasting and automatically handles seasonality, trend, and holiday effects without manual tuning. ARIMA is less automated; linear regression would require manual feature engineering for time components.

Full explanation →
80
Multi-Selecthard

Which THREE considerations are important when designing a batch prediction pipeline for a large dataset on Vertex AI?

Select 3 answers
A.Batch prediction automatically uses GPUs if the model framework requires them
B.Batch prediction requires a dedicated real-time endpoint
C.Choosing the appropriate machine type (e.g., n1-standard-16) balances cost and throughput
D.Large input files can be split into multiple smaller files to improve parallelism
E.Input data should be in Cloud Storage in a format supported by Vertex AI (e.g., JSONL, TFRecord)
AnswersC, D, E

Machine type impacts performance and cost.

Why this answer

Selecting the appropriate machine type, such as n1-standard-16, directly impacts the cost-performance trade-off in batch prediction. Vertex AI batch prediction jobs run on Compute Engine instances, and choosing a machine type with more vCPUs and memory can increase throughput for large datasets, but also raises cost. The key is to match the machine type to the model's computational needs and the data volume, avoiding over-provisioning while ensuring the job completes within acceptable time.

Exam trap

Google Cloud often tests the misconception that batch prediction requires a real-time endpoint or automatically uses GPUs, when in fact batch prediction is a serverless, endpoint-free process that requires explicit machine type and GPU configuration.

Full explanation →
81
MCQeasy

A data engineer needs to design a stream processing pipeline that reads events from Pub/Sub, enriches them with data from a Cloud Storage file, and writes aggregated results to BigQuery. The pipeline must handle late-arriving events up to 1 hour. Which Dataflow feature should be used to manage late data?

A.Triggers
B.Watermarks
C.Side inputs
D.Windowing
AnswerB

Watermarks track the event time progress and allowed lateness; Dataflow drops elements beyond the watermark.

Why this answer

Watermarks track event time progress and allow specifying allowed lateness. Triggers control when results are emitted, but watermarks handle late data.

Full explanation →
82
Multi-Selectmedium

You are designing a BigQuery data lake for a healthcare organization. The data includes patient records that must be access-controlled at the row level. Which TWO features should you use to meet this requirement?

Select 2 answers
A.Row-level security using row access policies
B.Authorised views with row filters
C.Dataset-level IAM roles
D.Clustering on patient_id
E.Materialised views
AnswersA, B

Row access policies filter rows based on user identity or group membership.

Why this answer

Row-level security in BigQuery allows filtering at the row level using access policies. Authorised views can also be used to expose only certain rows. Clustering and materialised views do not provide row-level access control.

Full explanation →
83
Multi-Selecthard

A company runs a Dataflow pipeline that processes a high-volume data stream. They notice that the pipeline's worker CPU utilisation is near 100% and the system lag is increasing. Which three actions can improve performance? (Choose three.)

Select 3 answers
A.Increase the worker disk size.
B.Increase the number of workers.
C.Use batch processing instead of streaming.
D.Enable Dataflow Streaming Engine.
E.Use higher-CPU machine types (e.g., n2-highcpu).
AnswersB, D, E

More workers distribute the load and reduce CPU per worker.

Why this answer

Increasing the number of workers distributes the processing load across more parallel workers, reducing CPU utilization per worker and allowing the pipeline to keep up with the incoming data stream. This directly addresses both high CPU usage and increasing system lag by scaling out horizontally.

Exam trap

A common trap is assuming that increasing disk size (Option A) improves CPU-related performance issues. In Dataflow, increasing disk size only helps with storage bottlenecks (e.g., shuffle disk overflow), not with high CPU utilization or system lag.

Full explanation →
84
MCQmedium

A financial company processes transactions in real-time and requires exactly-once processing semantics. They also need to reprocess historical data for backtesting. Which Google Cloud service should they use?

A.Cloud Pub/Sub
B.Cloud Functions
C.Cloud Dataproc
D.Cloud Dataflow
AnswerD

Supports exactly-once and batch/streaming.

Why this answer

Cloud Dataflow (D) is correct because it provides exactly-once processing semantics via its distributed snapshot mechanism (based on the MillWheel paper) and supports both real-time streaming and batch processing for historical backtesting under a unified programming model. This allows the company to reprocess historical data using the same pipeline code, ensuring consistency across real-time and batch modes.

Exam trap

Google Cloud often tests the misconception that Cloud Pub/Sub (A) provides exactly-once delivery, but in reality it offers at-least-once delivery, and candidates overlook Dataflow's unified batch/streaming model for reprocessing historical data.

How to eliminate wrong answers

Option A is wrong because Cloud Pub/Sub is a messaging service that offers at-least-once delivery by default, not exactly-once processing, and it lacks built-in capabilities for reprocessing historical data in a unified batch/streaming manner. Option B is wrong because Cloud Functions is an event-driven serverless compute service that does not provide exactly-once processing guarantees or native support for reprocessing large historical datasets; it is designed for lightweight, stateless functions. Option C is wrong because Cloud Dataproc is a managed Hadoop/Spark service that does not natively guarantee exactly-once processing semantics and requires manual handling of state and reprocessing logic, unlike Dataflow's automatic checkpointing.

Full explanation →
85
MCQmedium

What is the most likely cause of this error?

A.The BigQuery table is not partitioned
B.The Dataflow worker does not have the correct time zone
C.The pipeline is using a fixed window but the data is out of order
D.The schema of the BigQuery table expects a TIMESTAMP but the pipeline is sending a STRING
AnswerD

The error clearly shows an attempt to convert a string to a timestamp, indicating a schema mismatch.

Why this answer

The error message indicates a type mismatch: BigQuery expects a TIMESTAMP column, but the pipeline is sending a STRING. Dataflow's BigQuery sink performs automatic schema validation, and if the source data type (STRING) does not match the target column type (TIMESTAMP), the write operation fails with a mismatch error. This is a common issue when pipeline code or source data formats timestamps as strings without explicit conversion.

Exam trap

Google Cloud often tests the distinction between schema type mismatches and data ordering or partitioning issues, so candidates may confuse a type error with a windowing or time zone problem.

How to eliminate wrong answers

Option A is wrong because a non-partitioned BigQuery table would not cause a type mismatch error; it would instead cause performance issues or quota errors on large writes. Option B is wrong because the Dataflow worker's time zone affects timestamp interpretation, not the data type of the field being written; the error is about schema type, not time zone conversion. Option C is wrong because out-of-order data with a fixed window causes late data handling or watermark issues, not a schema type mismatch; the error is specifically about the data type sent to BigQuery.

Full explanation →
86
Multi-Selectmedium

Your streaming Dataflow pipeline reads from Pub/Sub and writes to BigQuery. You need to update the pipeline to add a new transformation step without losing any messages or causing duplicate processing. Which TWO actions should you take? (Choose 2)

Select 2 answers
A.Use the Dataflow update command with the same pipeline name and the new template.
B.Take a snapshot of the pipeline before updating.
C.Stop the pipeline, modify it, then restart with a new name.
D.Drain the pipeline before making changes.
E.Cancel the pipeline and create a new one.
AnswersA, B

Updating a running pipeline with the same name is the recommended way to apply changes without draining.

Why this answer

To update a streaming pipeline without draining, you can use the `--update` flag with the new pipeline. The snapshot feature allows you to restore state if needed. Draining would stop the pipeline; canceling would lose messages.

Full explanation →
87
MCQmedium

Your organization deploys multiple versions of the same model to Vertex AI Endpoint for A/B testing. You have a production model (v1) serving 90% of traffic and a candidate model (v2) serving 10%. After one week, you observe that v2 has a slightly lower AUC but significantly higher business metrics like click-through rate. The product team wants to gradually increase v2's traffic. However, you need to ensure that the overall prediction latency remains under 200 ms. Currently, the endpoint has 10 replicas for v1 and 2 replicas for v2. What is the best approach to roll out v2 while maintaining latency SLO?

A.Merge v2's model into v1 by retraining v1 with v2's architecture and deploy as a single model.
B.Immediately set v2 to serve 100% traffic and monitor latency; if it exceeds 200 ms, roll back.
C.Increase v2's traffic split by 10% each day while also adding replicas for v2 based on CPU utilization.
D.Use a separate endpoint for v2 and route traffic at the load balancer level.
AnswerC

Gradual increase with autoscaling ensures latency remains within bounds.

Full explanation →
88
MCQmedium

A data engineer needs to enforce that all datasets in a project expire after 90 days to reduce storage costs. They want to automate this without manual intervention. Which approach should they use?

A.Use Cloud Storage lifecycle rules to delete tables after 90 days
B.Create an IAM policy that revokes access after 90 days
C.Use BigQuery scheduled queries to delete tables older than 90 days
D.Set a default table expiration on each BigQuery dataset to 90 days
AnswerD

BigQuery dataset properties allow setting a default table expiration, automatically deleting tables after the specified days.

Why this answer

BigQuery datasets support a default table expiration setting that automatically deletes tables after a specified number of days. This enforces a 90-day lifecycle for all tables in the dataset without requiring manual intervention or external automation, directly addressing the cost reduction goal.

Exam trap

A common mistake is applying Cloud Storage lifecycle rules to BigQuery tables. BigQuery has its own table expiration settings at the dataset level, which are distinct from Cloud Storage object lifecycle management.

How to eliminate wrong answers

Option A is wrong because Cloud Storage lifecycle rules apply to objects in Cloud Storage buckets, not to BigQuery tables; BigQuery tables are stored separately and cannot be managed by Cloud Storage lifecycle policies. Option B is wrong because IAM policies control access permissions, not data lifecycle; revoking access does not delete tables or reduce storage costs. Option C is wrong because BigQuery scheduled queries can delete tables but require writing and maintaining custom SQL scripts and scheduling logic, introducing complexity and potential failure points, whereas a default table expiration is a declarative, server-managed setting that requires no ongoing maintenance.

Full explanation →
89
MCQmedium

Your team uses Looker Studio to build dashboards on top of BigQuery. The dashboards are slow when filtering on a high-cardinality dimension (e.g., user ID). You want to improve performance without changing the underlying BigQuery table design. Which action should you take?

A.Create a clustered table on the user ID column.
B.Apply a filter to limit data to the last month only.
C.Enable BigQuery BI Engine on the project.
D.Use Looker Studio's extract data functionality.
AnswerC

BI Engine accelerates queries by caching data in memory, improving Looker Studio dashboard performance.

Why this answer

BI Engine automatically caches the BigQuery tables in memory for Looker Studio, accelerating queries on any dimension. Enabling BI Engine on the project will improve dashboard performance.

Full explanation →
90
MCQmedium

Your team has implemented a CI/CD pipeline using Cloud Composer (Apache Airflow) to retrain a model every day. The pipeline reads new data from BigQuery, trains a model using Vertex AI Training, evaluates it, and if the accuracy improves, deploys it to a Vertex AI Endpoint. For the past week, the pipeline has been running successfully but no new model has been deployed because the evaluation accuracy never exceeds the previous model's accuracy. The training data volume has been consistent. You suspect that the model is not learning from the new data. What should you do?

A.Deploy the new model anyway and run an A/B test in production to see if it performs better online.
B.Examine the training data for any data quality issues such as missing values or label leakage.
C.Increase the training budget or number of training steps to allow the model to converge better.
D.Change the evaluation metric to a different one that may show improvement, such as F1 score instead of accuracy.
AnswerB

Data quality issues can prevent the model from learning meaningful patterns despite sufficient data volume.

Full explanation →
91
MCQmedium

An organization needs to continuously replicate change data from a MySQL database to BigQuery with sub-minute latency. The database is running on-premises. Which Google Cloud service should they use?

A.Cloud Pub/Sub with a custom connector to MySQL
B.BigQuery Data Transfer Service for MySQL
C.Cloud Dataflow with a JDBC source
D.Cloud Datastream
AnswerD

Datastream is purpose-built for CDC from common databases to BigQuery or GCS.

Why this answer

Datastream is a serverless change data capture (CDC) service that can replicate from MySQL, PostgreSQL, and Oracle to BigQuery or GCS with low latency. Pub/Sub is a messaging service and does not natively connect to MySQL. Dataflow can process streams but requires a CDC connector.

BigQuery Data Transfer Service does not support live CDC from on-prem MySQL.

Full explanation →
92
Matchingmedium

Match each data lifecycle stage to its description.

Drag a concept onto its matching description — or click a concept then click the description.

Concepts
Matches

Collecting data from various sources

Persisting data in a durable system

Transforming and analyzing data

Making data available for consumption

Moving data to long-term, low-cost storage

Why these pairings

Data collection and data storage are distinct stages. Collection is about obtaining raw data, while storage is about preserving it for future use. Common confusions arise due to the sequential nature of these processes.

Full explanation →
93
MCQeasy

You are monitoring a Dataflow streaming job and need to track the freshness of data being processed. What metric should you alert on?

A.Output throughput (elements/sec)
B.Error count
C.Data freshness (seconds)
D.CPU utilization
AnswerC

Data freshness measures the latency of the last processed event, indicating pipeline delay.

Why this answer

Data freshness (seconds) is the correct metric to alert on because it directly measures the lag between when an event occurs and when it is processed by the Dataflow pipeline. This metric, exposed as the 'system_lag' in Dataflow monitoring, indicates how up-to-date the output is relative to the input watermark. Alerting on data freshness ensures that the pipeline is meeting service-level agreements (SLAs) for real-time or near-real-time processing.

Exam trap

Google Cloud often tests the distinction between throughput and latency metrics, and the trap here is that candidates confuse high throughput with low latency, not realizing that a pipeline can process many elements per second while still having stale data due to watermark delays or unprocessed late data.

How to eliminate wrong answers

Option A is wrong because output throughput (elements/sec) measures processing rate, not timeliness; a pipeline can have high throughput but still be processing stale data due to backlog or watermark delays. Option B is wrong because error count tracks failures (e.g., exceptions, dropped elements) but does not indicate how current the processed data is; a pipeline with zero errors could still have high latency. Option D is wrong because CPU utilization is a resource metric that reflects compute efficiency, not data freshness; high CPU might cause delays, but it is an indirect indicator and not the direct measure of data staleness.

Full explanation →
94
MCQmedium

You have a BigQuery table that is partitioned by ingestion time and clustered on user_id. The table stores event logs and is queried frequently by user_id to analyze user behavior over the last 30 days. Queries are still scanning too many partitions. Which optimization should you apply first?

A.Create a materialized view that pre-aggregates data by user_id and date
B.Remove partitioning and rely solely on clustering
C.Change the partition column to a DATE column based on event_timestamp and keep clustering on user_id
D.Add clustering on a second column like event_type
AnswerC

If the ingestion time does not match the event timestamp, queries filtering on event time will not prune partitions effectively. Partitioning on the actual event date ensures partition pruning aligns with query filters.

Why this answer

The query filter on user_id already uses clustering, which prunes blocks within a partition. But the query also filters on a date range, which should leverage partition pruning. If queries still scan many partitions, the most likely cause is that the partition filter is not applied effectively.

Using a range filter on the partition column (_PARTITIONDATE) or a column used for partitioning will limit partitions scanned. But the question already says the table is partitioned by ingestion time. The best next step is to ensure the query uses a filter on the partition column.

However, among the options, changing the partition type to a specific date column (e.g., event_timestamp) with clustering on user_id could improve if the ingestion time doesn't align with the query time range.

Full explanation →
95
MCQhard

A company is migrating their on-premises Hadoop cluster to Google Cloud. The existing cluster runs HDFS, Hive, and Spark jobs. The migration must minimize changes to existing job code and configuration. The data volume is 50 TB and growing. The team expects to run both batch and interactive SQL queries. Which architecture should they use?

A.Keep HDFS on persistent Cloud Dataproc clusters and use BigQuery for SQL queries.
B.Use Cloud Dataflow for all batch processing and BigQuery for storage and querying.
C.Migrate HDFS to Cloud Storage, create a Cloud Dataproc cluster for Spark jobs, and use BigQuery for interactive SQL queries via a Hive metastore linked to BigQuery.
D.Use Cloud Dataproc with ephemeral clusters and Cloud Storage (instead of HDFS) for data storage. Run Spark jobs directly, and use Cloud Dataproc's built-in Hive on Cloud Dataproc for SQL queries.
AnswerD

Cloud Dataproc can use Cloud Storage as the data layer; most Spark and Hive jobs need minimal changes (e.g., file path prefix). Ephemeral clusters reduce cost. This preserves existing code.

Why this answer

It uses Cloud Storage as the underlying storage layer, which is HDFS-compatible and allows existing Spark jobs to run without code changes. Ephemeral Dataproc clusters reduce costs and provide native Hive support for interactive SQL queries, meeting both batch and interactive requirements without altering job configurations.

Exam trap

Google Cloud often tests the misconception that BigQuery must be used for all SQL queries in a migration, ignoring that Dataproc's Hive can directly query data in Cloud Storage without code changes, making it a simpler path for interactive SQL on existing Hive workloads.

How to eliminate wrong answers

Option A is wrong because keeping HDFS on persistent Dataproc clusters does not leverage Cloud Storage's scalability and cost benefits, and using BigQuery for SQL queries would require significant code changes to redirect queries away from Hive. Option B is wrong because Cloud Dataflow is not designed for Spark job compatibility, and using BigQuery for storage would break existing HDFS-based job code and configurations. Option C is wrong because linking a Hive metastore to BigQuery requires modifying the Hive configuration and does not support running Spark jobs directly on BigQuery storage without additional connectors, increasing complexity and potential code changes.

Full explanation →
96
Multi-Selecthard

Which TWO statements about designing a data processing pipeline on Google Cloud are correct? (Choose 2.)

Select 2 answers
A.Pub/Sub guarantees message ordering across all subscribers globally.
B.Cloud Bigtable is ideal for data warehousing and SQL analytics.
C.Dataproc is the best choice for fully managed data warehousing and analytics.
D.Cloud Data Fusion allows you to build and manage data pipelines visually without writing code.
E.Dataflow supports both batch and streaming modes in a single pipeline model.
AnswersD, E

Cloud Data Fusion provides a visual UI for designing pipelines.

Why this answer

Cloud Data Fusion provides a visual, no-code interface for building and managing data pipelines, enabling users to design ETL/ELT workflows through a drag-and-drop UI. It abstracts the underlying complexity of Apache Spark and Cloud Dataproc, making it suitable for users who prefer a graphical approach over writing code.

Exam trap

Google Cloud often tests the distinction between fully managed services (like BigQuery for warehousing) and managed cluster services (like Dataproc), as well as the limitations of Pub/Sub ordering guarantees, to see if candidates confuse operational databases with analytical systems.

Full explanation →
97
Multi-Selectmedium

A company wants to track data lineage for their BigQuery tables to understand how data flows from source to derived tables. Which TWO Google Cloud services can be used to capture and visualize data lineage? (Choose 2.)

Select 2 answers
A.Dataplex Data Lineage
B.Vertex AI Feature Store
C.Cloud Composer
D.Cloud Data Fusion
E.BigQuery Lineage API
AnswersA, E

Dataplex provides automated lineage tracking for BigQuery.

Why this answer

Dataplex provides a comprehensive Data Lineage feature that automatically captures lineage for BigQuery jobs. Additionally, the BigQuery Lineage API (part of Data Catalog) allows you to retrieve lineage programmatically.

Full explanation →
98
MCQeasy

A data engineer needs to design a batch pipeline that processes daily log files from Cloud Storage and writes aggregated results to BigQuery. Which service is most appropriate for this ETL job?

A.Cloud Pub/Sub with Cloud Functions
B.Cloud Composer
C.Cloud Data Fusion
D.Dataproc with PySpark
AnswerD

Dataproc handles large batch processing efficiently with Spark.

Why this answer

Dataproc with PySpark is the most appropriate choice because it provides a managed Spark/Hadoop environment that can efficiently process large daily log files stored in Cloud Storage using distributed computing. PySpark's native integration with BigQuery via the Spark BigQuery connector allows direct writing of aggregated results, making it ideal for batch ETL workloads that require complex transformations and high throughput.

Exam trap

The trap here is that candidates often confuse orchestration (Cloud Composer) with execution, or assume serverless options like Cloud Functions can handle heavy batch ETL, but the question specifically requires a service that performs the ETL processing, not just schedules or triggers it.

How to eliminate wrong answers

Option A is wrong because Cloud Pub/Sub with Cloud Functions is designed for event-driven, real-time streaming pipelines, not for batch processing of daily log files; Cloud Functions has a timeout limit (9 minutes for HTTP functions) and is not suited for heavy ETL jobs. Option B is wrong because Cloud Composer is a workflow orchestration tool (based on Apache Airflow) that schedules and monitors jobs, but it does not perform the actual data processing or transformation itself. Option C is wrong because Cloud Data Fusion is a visual data integration service for building pipelines, but it is more suited for low-code ETL and may lack the flexibility and performance of PySpark for large-scale batch log processing with custom transformations.

Full explanation →
99
Multi-Selectmedium

A data engineer needs to restrict access to BigQuery datasets such that only data from approved VPC networks can query them. They also need to audit data access. Which two security controls should they implement? (Choose two.)

Select 2 answers
A.Cloud Audit Logs
B.Customer-managed encryption keys (CMEK)
C.Data Loss Prevention (DLP)
D.IAM roles
E.VPC Service Controls
AnswersA, E

Audit logs record data access for auditing and monitoring.

Why this answer

Cloud Audit Logs (option A) is correct because it provides a record of all administrative and data-access operations on BigQuery datasets, enabling the data engineer to audit who accessed what data and from which network. This satisfies the requirement to audit data access by capturing detailed logs of API calls, including the identity of the caller and the source IP address.

Exam trap

Google Cloud often tests the distinction between identity-based controls (IAM) and network-based controls (VPC Service Controls), leading candidates to mistakenly choose IAM roles when the requirement explicitly specifies restricting access by VPC network rather than by user identity.

Full explanation →
100
MCQeasy

A data engineer needs to monitor model performance over time for drift detection. What tool is specifically designed for this?

A.Vertex AI Model Monitoring
B.Cloud Monitoring
C.Cloud Logging
D.BigQuery ML
AnswerA

Vertex AI Model Monitoring provides drift detection, skew detection, and alerts for deployed models.

Why this answer

Vertex AI Model Monitoring is specifically designed to detect prediction drift and feature skew in deployed machine learning models. It continuously analyzes serving data against training data distributions and alerts when statistical metrics (e.g., Jensen-Shannon divergence, L-infinity distance) exceed configured thresholds, making it the correct tool for drift detection in the context of operationalizing ML models.

Exam trap

Google Cloud often tests the distinction between general-purpose monitoring tools (Cloud Monitoring, Cloud Logging) and ML-specific monitoring services (Vertex AI Model Monitoring), trapping candidates who assume any monitoring tool can handle drift detection.

How to eliminate wrong answers

Option B (Cloud Monitoring) is wrong because it is a general-purpose infrastructure and application monitoring service for metrics, uptime, and alerting, not specialized for ML model drift detection. Option C (Cloud Logging) is wrong because it is a centralized log management and analysis service for storing and querying log data, not designed to compute statistical drift between training and serving distributions. Option D (BigQuery ML) is wrong because it is a service for creating and executing machine learning models using SQL queries in BigQuery, not a monitoring tool for detecting drift in already-deployed models.

Full explanation →
101
MCQeasy

An organization wants to ingest on-premises Oracle database changes into BigQuery for real-time analytics with minimal latency. The Oracle database is version 19c and has a high transactional volume. Which Google Cloud service should they use?

A.Datastream
B.Pub/Sub
C.Dataflow
D.Storage Transfer Service
AnswerA

Datastream supports real-time CDC from Oracle to BigQuery and GCS.

Why this answer

Datastream is the correct service because it is purpose-built for real-time, serverless change data capture (CDC) from Oracle databases (including 19c) to BigQuery. It uses LogMiner to read redo logs with minimal overhead, supporting high transactional volumes and sub-second latency without requiring custom code or manual schema management.

Exam trap

Candidates often incorrectly select Pub/Sub or Dataflow as generic streaming solutions, but Datastream is the only Google Cloud service that natively supports Oracle CDC without additional infrastructure.

How to eliminate wrong answers

Option B (Pub/Sub) is wrong because it is a messaging service for asynchronous event ingestion, not a CDC tool; it cannot directly read Oracle redo logs or handle schema evolution from a source database. Option C (Dataflow) is wrong because, while it can process streaming data, it requires a separate CDC connector (e.g., Debezium) and manual pipeline setup, adding complexity and latency compared to Datastream's managed Oracle-to-BigQuery integration. Option D (Storage Transfer Service) is wrong because it is designed for bulk file transfers (e.g., from on-premises NAS or S3) and cannot capture real-time database changes or connect to Oracle redo logs.

Full explanation →
102
MCQhard

A financial services company uses Cloud Bigtable to store trade data. They are experiencing hot-spotting on a single node, causing high latency. The row key format is [trade_id]#[timestamp]. Which row key design change would BEST distribute writes across tablets?

A.Use a hashed prefix of the trade_id, e.g., [hash(trade_id)]#[trade_id]#[timestamp]
B.Use a single row key of [timestamp]
C.Increase the number of Bigtable nodes to 20
D.Change row key to [timestamp]#[trade_id]
AnswerA

A hash prefix distributes writes across tablets by randomizing the start of the row key, reducing hot-spotting.

Why this answer

Adding a hashed prefix of the trade_id ensures that writes are evenly distributed across all Bigtable tablets. Bigtable partitions data by row key lexicographic order; without a hash, sequential trade IDs or timestamps cause all recent writes to land on a single tablet, creating a hotspot. The hash spreads the write load uniformly, regardless of the underlying key pattern.

Exam trap

A common trap in the Google Professional Data Engineer exam is confusing the order of key components. Simply reversing the order (e.g., putting the timestamp first) is often incorrectly thought to solve hot-spotting, but any monotonically increasing value at the start of the key will still cause a hotspot because Bigtable stores rows in lexicographic order.

How to eliminate wrong answers

Option B is wrong because using a single row key of [timestamp] would cause all writes with the same timestamp to collide on one row, creating an extreme hotspot and violating Bigtable's requirement for unique, distributed row keys. Option C is wrong because increasing the number of nodes does not fix a row key design flaw; Bigtable cannot rebalance writes if the row key pattern forces all traffic to a single tablet, and adding nodes only helps if the load is already distributed. Option D is wrong because reversing the order to [timestamp]#[trade_id] still places all writes with the same timestamp adjacent in lexicographic order, so recent timestamps will still hotspot on a single tablet; it does not introduce the randomness needed for distribution.

Full explanation →
103
MCQhard

A healthcare organization is deploying a model that processes protected health information (PHI). They need to ensure that the inference data is encrypted in transit and at rest, and access is audited. Which combination of services meets these requirements?

A.Cloud Run with VPC connector and Cloud KMS
B.Vertex AI Endpoints with IAM and Cloud Monitoring
C.Vertex AI Endpoints with VPC-SC, CMEK, and Cloud Audit Logs
D.AI Platform Prediction with Cloud Armor
AnswerC

VPC Service Controls protect against unauthorized data movement, CMEK for customer-managed encryption keys, and Cloud Audit Logs for compliance.

Why this answer

VPC Service Controls (VPC-SC) provides data exfiltration protection and ensures inference data remains within a defined security perimeter, Customer-Managed Encryption Keys (CMEK) encrypt data at rest with keys controlled by the organization, and Cloud Audit Logs capture all access events for auditing. This combination directly addresses encryption in transit (via VPC-SC perimeter enforcement) and at rest (via CMEK), plus access auditing via Cloud Audit Logs.

Exam trap

Google Cloud often tests the misconception that IAM and Cloud Monitoring alone satisfy encryption and auditing requirements, but IAM controls access without encrypting data at rest, and Cloud Monitoring tracks performance metrics, not access logs; candidates must recognize that VPC-SC, CMEK, and Cloud Audit Logs are the specific services needed for encryption and auditing in ML inference.

How to eliminate wrong answers

Option A is wrong because Cloud Run with VPC connector and Cloud KMS does not provide a managed inference endpoint optimized for ML models; Cloud Run is a general-purpose compute service, and while VPC connector enables private networking and Cloud KMS manages encryption keys, it lacks the specific model hosting, scaling, and monitoring capabilities of Vertex AI Endpoints. Option B is wrong because Vertex AI Endpoints with IAM and Cloud Monitoring provides access control and performance monitoring but does not encrypt data at rest with customer-managed keys (CMEK) or enforce a data perimeter via VPC-SC; Cloud Monitoring logs metrics, not access events for auditing. Option D is wrong because AI Platform Prediction (now legacy) with Cloud Armor provides DDoS protection but does not encrypt data at rest with CMEK or provide VPC-SC perimeter controls; Cloud Armor operates at the network edge and does not address encryption or audit logging requirements.

Full explanation →
104
Multi-Selectmedium

You are designing a data pipeline for a financial services company that requires exactly-once processing semantics. Which TWO services or configurations provide exactly-once guarantees?

Select 2 answers
A.Pub/Sub with exactly-once delivery enabled
B.Dataproc with checkpointing
C.Dataflow with exactly-once processing mode
D.Pub/Sub Lite with at-least-once delivery
E.Cloud Storage with object versioning
AnswersA, C

Pub/Sub offers exactly-once delivery for pull subscriptions when enabled.

Why this answer

Pub/Sub with exactly-once delivery enabled provides exactly-once message delivery. Dataflow with exactly-once processing mode ensures each record is processed exactly once. Pub/Sub Lite and Dataproc do not provide exactly-once guarantees, and Cloud Storage is for storage.

Full explanation →
105
MCQmedium

An application requires a globally distributed, strongly consistent database with 99.999% availability SLA. The workload is OLTP with high throughput across continents. Which service fits best?

A.Cloud SQL with cross-region replicas
B.Firestore
C.Cloud Bigtable
D.Cloud Spanner
AnswerD

Correct: global distribution, strong consistency, 99.999% SLA.

Why this answer

Cloud Spanner is the only service that provides globally distributed, strongly consistent (external consistency via TrueTime) OLTP with 99.999% availability SLA. It supports high-throughput ACID transactions across continents using synchronous replication and atomic clocks, meeting all stated requirements.

Exam trap

Candidates often confuse Cloud SQL cross-region replicas as providing strong consistency, but those replicas are eventually consistent due to asynchronous replication, and they do not meet the 99.999% SLA.

How to eliminate wrong answers

Option A is wrong because Cloud SQL with cross-region replicas uses asynchronous replication, which cannot guarantee strong consistency across regions and offers only a 99.95% SLA for regional instances, not 99.999%. Option B is wrong because Firestore is a NoSQL document database that provides strong consistency only within a single region; its multi-region mode uses eventual consistency for global reads, and it lacks the ACID transaction support needed for high-throughput OLTP across continents. Option C is wrong because Cloud Bigtable is a wide-column NoSQL database designed for analytical workloads with high throughput, but it does not support SQL queries, ACID transactions, or strong consistency across regions; it offers only single-row transactions and eventual consistency for multi-cluster replication.

Full explanation →
106
MCQmedium

A company wants to run hybrid transactional and analytical workloads on a PostgreSQL-compatible database with high performance. Which service should they choose?

A.Cloud Spanner
B.Cloud SQL for PostgreSQL
C.BigQuery
D.AlloyDB
AnswerD

Correct: AlloyDB has a columnar engine for analytics and is PostgreSQL-compatible.

Why this answer

AlloyDB is the correct choice because it is a fully managed PostgreSQL-compatible database service specifically designed for high-performance hybrid transactional and analytical workloads. It combines transactional processing with built-in columnar analytics, delivering up to 4x faster transactional performance and up to 100x faster analytical queries than standard PostgreSQL, without requiring any schema changes or ETL.

Exam trap

Google often tests the distinction between 'PostgreSQL-compatible' and 'PostgreSQL-based' — candidates mistakenly choose Cloud SQL for PostgreSQL because it is a managed PostgreSQL service, but they overlook the specific requirement for hybrid transactional and analytical workloads, which AlloyDB uniquely addresses with its integrated columnar engine.

How to eliminate wrong answers

Option A is wrong because Cloud Spanner is a globally distributed, strongly consistent relational database that is not PostgreSQL-compatible (it uses GoogleSQL or standard SQL with Spanner-specific extensions) and is optimized for horizontal scaling across regions, not for hybrid transactional/analytical workloads with PostgreSQL compatibility. Option B is wrong because Cloud SQL for PostgreSQL is a fully managed PostgreSQL service but is designed primarily for transactional (OLTP) workloads and lacks the built-in columnar engine and analytical acceleration needed for hybrid workloads, resulting in significantly slower analytical queries. Option C is wrong because BigQuery is a serverless, highly scalable data warehouse for analytical (OLAP) workloads, not a transactional database, and it is not PostgreSQL-compatible (it uses BigQuery SQL).

Full explanation →
107
MCQmedium

A company needs to store petabytes of time-series IoT sensor data and query it with single-digit millisecond latency at millions of reads per second. The data has a simple key-value structure with timestamps. Which Google Cloud database is MOST appropriate?

A.BigQuery
B.Firestore
C.Cloud Bigtable
D.Cloud Spanner
AnswerC

Bigtable is the correct choice: wide-column NoSQL, designed for time-series and IoT workloads, single-digit ms latency, and scales to millions of QPS with additional nodes.

Why this answer

Cloud Bigtable is the correct choice because it is a fully managed, scalable NoSQL database designed for large analytical and operational workloads, offering consistent sub-10ms latency for high-throughput reads and writes. It natively supports time-series data with row key design optimized for timestamp-based queries, and can handle millions of reads per second across petabytes of data, making it ideal for IoT sensor data.

Exam trap

A common pitfall is assuming BigQuery is suitable for real-time, high-throughput key-value lookups because of its speed on analytical queries, but BigQuery is not designed for point reads at millions of operations per second with single-digit millisecond latency.

How to eliminate wrong answers

Option A is wrong because BigQuery is a serverless data warehouse optimized for complex analytical SQL queries on large datasets, not for single-digit millisecond point lookups at millions of reads per second; its latency is typically in the hundreds of milliseconds to seconds. Option B is wrong because Firestore is a mobile/document database designed for real-time sync and moderate throughput, not for petabyte-scale time-series data with millions of reads per second; it has throughput limits and higher latency for such workloads. Option D is wrong because Cloud Spanner is a globally distributed relational database with strong consistency and SQL support, but it is overkill for simple key-value time-series data and incurs higher latency and cost compared to Bigtable for this specific use case.

Full explanation →
108
MCQeasy

A data engineer needs to design a data pipeline that ingests streaming data from Cloud Pub/Sub, performs real-time aggregations, and loads the results into BigQuery for dashboarding. Which Google Cloud service should they use for the streaming aggregation step?

A.Cloud Functions
B.Dataflow
C.Cloud Dataproc
D.Cloud Composer
AnswerB

Dataflow is designed for streaming and batch pipelines, with native Pub/Sub and BigQuery IOs.

Why this answer

Dataflow is a fully managed service for stream and batch processing that integrates with Pub/Sub and BigQuery. It supports exactly-once processing and low-latency streaming.

Full explanation →
109
Multi-Selectmedium

You are optimizing a Dataflow pipeline that performs a group-by-key transformation on a large, skewed dataset. The pipeline is experiencing high latency due to data skew (some keys have many more values). Which TWO actions can help mitigate the skew? (Choose two.)

Select 2 answers
A.Use hot key detection and split the hot key into multiple sub-keys (e.g., append a random number).
B.Enable the Dataflow service's automatic reshuffling feature.
C.Use CoGroupByKey to reduce the number of keys.
D.Increase the number of worker machines.
E.Use Combine.perKey with a combiner to aggregate values locally before shuffling.
AnswersA, E

Splitting a hot key distributes its values across multiple workers, reducing bottleneck.

Why this answer

Splitting a hot key into multiple sub-keys (e.g., by appending a random number) distributes the values across multiple shards during the shuffle phase, reducing the load on any single worker. This technique, often called "salting," is a standard pattern in Dataflow and Apache Beam to handle data skew by breaking the bottleneck caused by a single key with disproportionately many values.

Exam trap

Google Cloud often tests the misconception that simply adding more workers (Option D) or enabling automatic reshuffling (Option B) can fix data skew, when in fact these actions do not address the root cause of a single key being processed by one shard.

Full explanation →
110
MCQhard

Your MLOps pipeline uses Vertex AI Pipelines. You want to ensure that model training uses a consistent environment with specific Python package versions. Which approach best achieves this?

A.Include a requirements.txt file in the pipeline step and let Vertex AI install them
B.Use a pre-built deep learning container from Deep Learning Containers and install packages at runtime
C.Specify the Python version and package versions in the training job configuration
D.Build a custom container image with all dependencies and use it in the training step
AnswerD

Custom containers ensure exact same environment.

Why this answer

Building a custom container image with all dependencies ensures a fully deterministic and reproducible environment for model training. Vertex AI Pipelines executes each step as a container, so by pre-installing specific Python package versions into a custom image, you eliminate any risk of version drift or network issues during package installation at runtime. This approach aligns with MLOps best practices for environment consistency and is the most reliable method when exact package versions are critical.

Exam trap

Google Cloud often tests the distinction between runtime configuration (options A, B, C) and pre-built containerization (option D), trapping candidates who think specifying versions in a config file or installing at runtime is sufficient for full environment consistency in a pipeline context.

How to eliminate wrong answers

Option A is wrong because including a requirements.txt file and letting Vertex AI install them at runtime introduces variability; the installation may fail due to network issues, dependency conflicts, or changes in package repositories, and it does not guarantee the same environment across pipeline retries. Option B is wrong because using a pre-built deep learning container and installing packages at runtime still relies on runtime installation, which can lead to inconsistent environments if the installation process fails or if package versions are not pinned correctly. Option C is wrong because specifying Python version and package versions in the training job configuration only applies to AI Platform Training jobs, not to Vertex AI Pipelines; Vertex AI Pipelines runs steps as containers and does not natively support specifying package versions in the pipeline step configuration—the environment must be defined within the container image.

Full explanation →
111
MCQhard

A company uses BigQuery and wants to reduce query costs by using BI Engine for Looker Studio dashboards. The data is stored in a BigQuery dataset with 5 TB of frequently accessed tables. The dashboards run dozens of concurrent queries. What is the recommended approach to enable BI Engine acceleration?

A.Enable BI Engine by setting the dataset option 'enable_bi_engine=TRUE' in the dataset metadata.
B.Grant all Looker Studio users the 'biengine.user' IAM role on the project.
C.Create a reservation in the Administration panel and assign it to the project.
D.Create materialized views of the tables and connect Looker Studio to the views.
AnswerC

BI Engine requires creating a capacity reservation (memory) in the region where your data resides. This reservation automatically accelerates queries from Looker Studio.

Why this answer

BI Engine is a reserved capacity service that caches data in memory. You must reserve capacity (amount of memory) for a specific BigQuery region, and then BI Engine automatically accelerates queries from Looker Studio and other BI tools. It does not require you to grant specific IAM roles to users (they just need BigQuery permissions) or to create materialized views.

You do not need to enable it per dataset; it works at the project/region level.

Full explanation →
112
MCQmedium

Refer to the exhibit. An ML engineer sees this error when invoking a Vertex AI endpoint. What is the most likely cause?

A.The input format should be JSON
B.The model has a bug in the ResNet50 architecture
C.The model expects 128x128 images but raw input is 256x256
D.The endpoint is overloaded
AnswerC

The error shows expected shape [1,128,128,3] but got [1,256,256,3], indicating image size mismatch.

Why this answer

The error indicates a mismatch between the input dimensions expected by the model and the dimensions of the data being sent to the Vertex AI endpoint. ResNet50 models are commonly trained on 128x128 images, and if the raw input is 256x256, the endpoint will reject the request because the model's input tensor shape does not match. This is a typical input validation error in Vertex AI, where the serving infrastructure checks the shape of the prediction request against the model's signature.

Exam trap

Google Cloud often tests the distinction between input validation errors (e.g., shape mismatch) and model logic errors (e.g., architecture bugs), so candidates mistakenly attribute the error to a model bug or endpoint overload rather than a simple data preprocessing mismatch.

How to eliminate wrong answers

Option A is wrong because the error is about image dimensions, not the serialization format; Vertex AI endpoints accept JSON by default, and the error message would explicitly mention 'invalid format' if that were the issue. Option B is wrong because a bug in the ResNet50 architecture would cause inference errors or incorrect predictions, not a dimension mismatch error at the endpoint level. Option D is wrong because an overloaded endpoint would return a 429 HTTP status code or a 'resource exhausted' error, not a dimension mismatch error.

Full explanation →
113
MCQeasy

You want to quickly estimate the number of distinct visitors to your website from a large BigQuery table. Which function provides an approximate count with low latency?

A.APPROX_COUNT_DISTINCT
B.HyperLogLog++
C.COUNT(DISTINCT)
D.APPROX_QUANTILES
AnswerA

Approximate count with low latency.

Why this answer

APPROX_COUNT_DISTINCT provides an approximate distinct count with low latency. COUNT(DISTINCT) is exact but slower for large data. APPROX_QUANTILES estimates quantiles.

HyperLogLog is not a BQ function.

Full explanation →
114
MCQeasy

Your company runs batch predictions using Vertex AI Batch Prediction on a monthly basis. The predictions are used to generate customer segments for marketing campaigns. This month, the batch prediction job failed with an error: 'The number of rows in the input table does not match the number of rows in the output table.' The input table in BigQuery has 5 million rows, but the output table has only 4.5 million rows. You need to identify and handle the missing predictions. What is the most efficient course of action?

A.Manually inspect the input table to find which rows are missing and rerun the batch prediction for those rows.
B.Run the batch prediction job with the 'generate_explanation' parameter enabled to get additional output for debugging.
C.Enable the 'write_prediction_errors' flag in the batch prediction configuration to capture failed predictions in a separate table.
D.Use a Cloud Dataflow pipeline to process the input data and call the model for each row, handling errors programmatically.
AnswerC

This flag causes failed predictions to be written to an error table, allowing you to identify and correct the problematic rows.

Full explanation →
115
MCQeasy

You are using AI Platform Prediction (now Vertex AI) for online predictions. You notice that some requests are failing with a 503 status code. Which is the most likely cause?

A.The model is experiencing high traffic and the underlying nodes are still scaling up
B.The input data format does not match the model's expected schema
C.The project has exceeded its prediction requests quota
D.The service account used for prediction does not have the required permissions
AnswerA

503 errors often occur during scaling.

Why this answer

A 503 status code in Vertex AI (formerly AI Platform Prediction) indicates that the prediction service is temporarily unavailable, most commonly due to autoscaling latency. When a model receives a sudden spike in traffic, the underlying nodes (compute instances) may still be provisioning and initializing, causing requests to be rejected until the new nodes are ready to serve. This is a transient condition that resolves once scaling completes.

Exam trap

Google Cloud often tests the distinction between HTTP 503 (service unavailable, transient) and HTTP 429 (quota exceeded) or HTTP 400 (bad request), so candidates mistakenly attribute scaling issues to quota exhaustion or permission errors.

How to eliminate wrong answers

Option B is wrong because a mismatch in input data format (e.g., wrong tensor shape or feature names) would result in a 400 Bad Request error, not a 503. Option C is wrong because exceeding prediction request quota would return a 429 Too Many Requests error, not a 503. Option D is wrong because insufficient permissions (e.g., missing `aiplatform.predict` role) would cause a 403 Forbidden error, not a 503.

Full explanation →
116
MCQhard

A company processes large volumes of GPS sensor data stored in Cloud Storage. Each hour, they run an Apache Spark job that aggregates the data by geohash region. The job must be cost-effective and scale automatically. Currently, they are using a Dataproc cluster with preemptible workers. Which improvement would best reduce costs while maintaining performance?

A.Use a larger Dataproc cluster with standard workers
B.Migrate the job to BigQuery scheduled queries
C.Switch to Dataflow batch pipeline with Apache Beam
D.Use Dataproc Serverless Spark
AnswerD

Dataproc Serverless Spark runs Spark jobs without cluster management, scales automatically, and you pay only for resources used, reducing cost.

Why this answer

Dataproc Serverless Spark (Option D) eliminates the need to manage a cluster, automatically scaling resources to match job demand and charging only for the resources consumed during execution. This removes the overhead of preemptible worker management and idle cluster costs, directly reducing expenses while maintaining performance for the hourly aggregation job.

Exam trap

Google Cloud often tests the misconception that migrating to a different processing engine (like Dataflow or BigQuery) is always the best cost-saving move, when in fact reusing existing Spark code on a serverless platform avoids migration costs and leverages the same API.

How to eliminate wrong answers

Option A is wrong because using a larger cluster with standard workers increases costs due to higher per-hour instance pricing and potential idle time, without addressing the cost inefficiency of preemptible workers. Option B is wrong because BigQuery scheduled queries are designed for SQL-based analytics on data already in BigQuery, not for processing large volumes of GPS sensor data stored in Cloud Storage with Apache Spark aggregations; migrating would require rewriting the Spark logic and may incur high BigQuery slot costs. Option C is wrong because while Dataflow batch pipelines with Apache Beam can process data cost-effectively, they require rewriting the existing Spark job into Beam, introducing development overhead and potential performance differences, whereas Dataproc Serverless Spark directly runs the existing Spark code without migration.

Full explanation →
117
MCQeasy

A company needs to stream real-time user click events from a web application to BigQuery for analysis. Which Google Cloud architecture is most suitable?

A.App Engine -> Pub/Sub -> Dataflow -> BigQuery
B.Cloud Scheduler -> BigQuery
C.Compute Engine -> Cloud Storage -> BigQuery
D.Cloud Functions -> BigQuery
AnswerA

This architecture supports real-time streaming with decoupled components.

Why this answer

It provides a fully managed, scalable, and decoupled architecture for ingesting real-time click events. Pub/Sub acts as a durable, asynchronous message buffer that can handle high-throughput streams, Dataflow (Apache Beam) processes the events in near real-time with exactly-once semantics, and BigQuery serves as the analytics warehouse. This pattern is the recommended Google Cloud approach for streaming analytics, as it decouples producers from consumers and supports auto-scaling.

Exam trap

The trap here is that candidates often choose Cloud Functions (Option D) thinking it is sufficient for real-time ingestion, but they overlook its execution timeout and lack of built-in streaming semantics, which makes it unsuitable for sustained high-throughput event pipelines.

How to eliminate wrong answers

Option B is wrong because Cloud Scheduler is a cron job service for triggering actions on a schedule, not a real-time event ingestion mechanism; it cannot stream continuous click events. Option C is wrong because Compute Engine and Cloud Storage are batch-oriented; writing events directly to Cloud Storage introduces latency and requires additional batch processing to load into BigQuery, making it unsuitable for real-time streaming. Option D is wrong because Cloud Functions has a 9-minute timeout and is designed for short-lived, event-driven compute, not for continuous, high-throughput streaming; it would also require custom code to buffer and batch writes to BigQuery, losing the managed streaming capabilities of Dataflow.

Full explanation →
118
MCQeasy

A data engineer runs this Dataflow template to load CSV files from Cloud Storage into BigQuery. The job fails with a 'File pattern not matching any files' error. What is the most likely cause?

A.The bucket name is incorrectly spelled
B.The CSV files are stored in a subdirectory that is not matched by the pattern
C.The template has a bug
D.The output table does not exist
AnswerB

The pattern `*.csv` in a prefix does not include files in nested subdirectories.

Why this answer

The error 'File pattern not matching any files' indicates that the file pattern specified in the Dataflow template does not resolve to any existing objects in Cloud Storage. If the CSV files are stored in a subdirectory (e.g., gs://bucket/subdir/*.csv) but the pattern only references the root (e.g., gs://bucket/*.csv), no files will be matched. This is the most likely cause because the pattern must explicitly include the subdirectory path.

Exam trap

Google Cloud often tests the distinction between file pattern matching errors and bucket-level errors, trapping candidates who confuse a missing subdirectory in the pattern with a misspelled bucket name.

How to eliminate wrong answers

Option A is wrong because an incorrectly spelled bucket name would result in a 'bucket not found' or 'access denied' error, not a 'file pattern not matching any files' error. Option C is wrong because the template is a well-tested Google-provided template; a bug is unlikely and would typically cause different errors (e.g., runtime exceptions). Option D is wrong because the output table not existing would cause a BigQuery table creation or write error, not a file pattern matching error in Cloud Storage.

Full explanation →
119
MCQeasy

You need to schedule a Dataproc Spark job to run at 2 AM every day, and upon completion, trigger a BigQuery load job. Which Cloud Composer operator should you use to run the Spark job?

A.DataflowPythonOperator
B.BigQueryOperator
C.DataprocClusterCreateOperator
D.DataprocSubmitJobOperator
AnswerD

This operator submits a job (e.g., Spark, PySpark) to an existing Dataproc cluster.

Why this answer

The DataprocSubmitJobOperator is specifically designed to submit a job (e.g., a Spark job) to an existing Dataproc cluster. In this scenario, you need to run a Spark job on a scheduled basis, and Cloud Composer (Airflow) provides this operator to submit the job to Dataproc. After the Spark job completes, you can chain a BigQuery load operator to trigger the load, matching the requirement exactly.

Exam trap

The trap here is that candidates confuse operators that manage cluster lifecycle (like DataprocClusterCreateOperator) with operators that submit jobs, or they mistakenly think DataflowPythonOperator can run Spark jobs because both are data processing frameworks.

How to eliminate wrong answers

Option A is wrong because DataflowPythonOperator is used to run Apache Beam pipelines on Dataflow, not Spark jobs on Dataproc. Option B is wrong because BigQueryOperator is used to execute BigQuery SQL queries or load jobs, not to run Spark jobs. Option C is wrong because DataprocClusterCreateOperator is used to create a new Dataproc cluster, not to submit a job to an existing cluster; the question assumes the cluster already exists or is managed separately, and the focus is on submitting the Spark job.

Full explanation →
120
MCQhard

A company uses Cloud Pub/Sub to ingest events from multiple sources. They need to guarantee that each event is processed exactly once by downstream consumers. However, Pub/Sub guarantees at-least-once delivery. Which additional steps should they implement to achieve exactly-once processing?

A.Set the subscription's acknowledgment deadline to 0.
B.Enable message deduplication on the subscription.
C.Store each message's unique ID in a database and ignore duplicates.
D.Use a dead letter topic to capture duplicates.
AnswerC

Consumer-side deduplication by tracking message IDs achieves exactly-once processing.

Why this answer

Pub/Sub only provides at-least-once; exactly-once requires consumer-side deduplication using a unique message ID.

Full explanation →
121
MCQhard

A team configured a garbage collection rule on a Cloud Bigtable column family with max_age of 100 seconds. After 2 minutes, they notice that data older than 100 seconds is still present. What is the most likely reason?

A.They need to apply the rule using a different API
B.Garbage collection runs only periodically (e.g., once per day)
C.The max_age must be at least 1 hour
D.The rule is applied only to new data, not existing data
AnswerB

Bigtable GC runs in the background at intervals (by default once per day), so newly set rules may not take effect immediately.

Why this answer

Cloud Bigtable garbage collection (GC) is not applied in real time; it runs as a background process that typically executes once per day. Even though the max_age rule is set to 100 seconds, the actual deletion of expired data occurs only during the next scheduled GC cycle, which may not happen for up to 24 hours. Therefore, observing data older than 100 seconds after only 2 minutes is expected behavior.

Exam trap

The trap here is that candidates assume garbage collection is immediate or near-real-time, but Cloud Bigtable's GC is a batch process with a long interval (typically daily), so data persists until the next scheduled run.

How to eliminate wrong answers

Option A is wrong because Cloud Bigtable garbage collection rules are configured via the standard Cloud Bigtable API (e.g., gcloud bigtable instances tables update or the client library's modify_column_family method); no different API is required. Option C is wrong because Cloud Bigtable does not enforce a minimum max_age of 1 hour; the max_age can be set to any positive duration, including 100 seconds. Option D is wrong because garbage collection rules apply to both existing and new data; the rule is not limited to new data only—it governs all data in the column family once the rule is set.

Full explanation →
122
MCQeasy

You need to track data lineage from a BigQuery table through a series of transformations and into a Vertex AI model training pipeline. Which Google Cloud service provides automated data lineage tracking?

A.Dataplex
B.Dataflow
C.Data Catalog
D.Cloud Composer
AnswerA

Dataplex provides automated data lineage tracking for BigQuery and Vertex AI pipelines.

Why this answer

Dataplex includes data lineage tracking that captures metadata about data movement and transformations across BigQuery, Vertex AI, and other services. Data Catalog provides metadata but not automated lineage.

Full explanation →
123
MCQeasy

Which BigQuery function can be used to retrieve the value of a column from the previous row within a partition, ordered by a timestamp?

A.LAG()
B.FIRST_VALUE()
C.ROW_NUMBER()
D.LEAD()
AnswerA

Correct: LAG() accesses data from a previous row.

Why this answer

LAG() is a window function that returns the value of a column from a row that is a specified number of rows before the current row within the partition. LEAD() retrieves from a following row.

Full explanation →
124
Multi-Selecthard

A data engineering team is designing a streaming pipeline using Cloud Dataflow. They need to join two unbounded PCollections based on a common key. The join must handle late data up to 10 minutes. Which THREE components should they use?

Select 3 answers
A.CoGroupByKey transform
B.Window into fixed windows
C.Use a global window with triggers
D.Flatten transform
E.Set allowed lateness and trigger to handle late data
AnswersA, B, E

CoGroupByKey performs a join of two PCollections by key.

Why this answer

CoGroupByKey joins two PCollections by key. Window into fixed windows of appropriate duration. Allowed lateness and triggers handle late data.

Full explanation →
125
MCQeasy

A company is using Looker to explore their BigQuery data. They have defined a LookML model with an 'explore' that joins two views: 'orders' and 'customers'. The join is a left join. They want to ensure that only customers with orders are shown when exploring. Which LookML parameter should they modify?

A.Set the 'required_joins' parameter on the 'orders' view
B.Use a derived table with a WHERE clause
C.Add a filter in the 'customers' view to exclude nulls
D.Change the join type from 'left_outer' to 'inner'
AnswerD

Inner join will only include customers that have at least one order.

Why this answer

The 'join_type' parameter in LookML defines the join type. Changing it to 'inner' will exclude customers without orders.

Full explanation →
126
Multi-Selecthard

A company runs BigQuery workloads with varying demand. They want to use flat-rate pricing with baseline slots and the ability to burst during peak times. Which TWO actions should they take? (Choose 2)

Select 2 answers
A.Use on-demand pricing for bursting
B.Use flex slots for short-term bursts
C.Set a maximum number of slots per query
D.Purchase committed use reservations for baseline slots
E.Create a reservation with baseline + autoscaling slots
AnswersB, D

Flex slots are ideal for handling peak demand without commitment.

Why this answer

To achieve flat-rate pricing with baseline slots and bursting, purchase committed use reservations for the baseline (option D) to get a discount on consistent usage, and use flex slots for short-term bursts (option B) which are also flat-rate. Option E is incorrect because autoscaling with reservations can use idle committed slots for bursts, but it may still fall back to on-demand pricing if no idle slots are available, so it does not ensure that all bursts are covered by flat-rate pricing. Flex slots (option B) guarantee flat-rate bursting.

Full explanation →
127
MCQeasy

A data scientist wants to test a new model version on a small percentage of traffic before full rollout. Which Vertex AI feature allows this?

A.A/B testing
B.Endpoint traffic splitting
C.Model monitoring
D.Model versioning with canary deployments
AnswerB

Traffic splitting allows routing a subset of requests to a different model version.

Why this answer

Vertex AI Endpoint traffic splitting allows you to route a specified percentage of inference requests to different model versions deployed on the same endpoint. This enables gradual rollout by directing a small fraction of traffic (e.g., 5%) to the new model while the rest goes to the current version, without needing separate endpoints or manual routing logic.

Exam trap

The trap here is that candidates confuse the conceptual practice of 'canary deployments' (Option D) with the specific Vertex AI feature 'endpoint traffic splitting' (Option B), but the exam expects the exact feature name as defined in the Google Cloud documentation.

How to eliminate wrong answers

Option A is wrong because A/B testing in Vertex AI is a feature for comparing model performance metrics (like accuracy or latency) by splitting traffic, but it is not the feature that directly enables traffic splitting itself—traffic splitting is the underlying mechanism, and A/B testing is a higher-level evaluation tool built on top of it. Option C is wrong because Model monitoring is used to detect data drift, feature skew, and prediction anomalies on deployed models, not to control traffic distribution between versions. Option D is wrong because model versioning with canary deployments is a conceptual practice, not a specific Vertex AI feature; the actual feature that implements canary-style traffic routing is endpoint traffic splitting, which is the correct answer.

Full explanation →
128
Multi-Selecteasy

A data engineer needs to schedule a recurring transfer of data from a partner's Amazon S3 bucket to a Cloud Storage bucket for further processing. Which THREE components or configurations are necessary? (Choose 3)

Select 3 answers
A.A VPC network configuration
B.Specification of the source S3 bucket and destination GCS bucket
C.A scheduled transfer job in Storage Transfer Service
D.A Pub/Sub topic to notify completion
E.Authentication credentials for AWS (e.g., access key and secret)
AnswersB, C, E

Source and destination are required.

Why this answer

Scheduling, source and destination locations, and authentication/authorization are essential for a Storage Transfer Service job.

Full explanation →
129
MCQmedium

A retail company uses a Vertex AI endpoint to serve product recommendations. The model is a TensorFlow model deployed with a custom container. Recently, users have reported that recommendations are stale. The model is retrained daily using Vertex AI Pipelines. The pipeline completes successfully, but the endpoint continues to serve the old model. The team checks the pipeline logs and sees that the new model is uploaded to the Vertex AI Model Registry. The endpoint has traffic split set to 100% for the old model. The team needs to update the endpoint to serve the new model version. What should they do?

A.Check the pipeline for errors in the deployment step
B.Re-upload the model with a different version ID
C.Redeploy the same model to the endpoint
D.Update the endpoint to deploy the new model version from the registry and adjust traffic split
AnswerD

Explicitly deploy new version to endpoint.

Why this answer

The pipeline successfully uploaded the new model to the Vertex AI Model Registry, but the endpoint still has its traffic split configured to 100% for the old model. To serve the new model, the team must explicitly update the endpoint to deploy the new model version from the registry and adjust the traffic split to route 100% of traffic to it. This is a standard operational step in Vertex AI: uploading a model does not automatically update the endpoint's deployment or traffic allocation.

Exam trap

Google Cloud often tests the misconception that uploading a new model version to the registry automatically updates the endpoint's serving configuration, when in fact the traffic split must be explicitly adjusted to route requests to the new model.

How to eliminate wrong answers

Option A is wrong because the pipeline logs show no errors in the deployment step; the model was successfully uploaded to the registry, so checking for errors is unnecessary and misdiagnoses the issue. Option B is wrong because re-uploading the model with a different version ID does not change the endpoint's deployment or traffic split; the endpoint still points to the old model version. Option C is wrong because redeploying the same model (the old version) to the endpoint would not serve the new model; the team needs to deploy the new model version from the registry, not redeploy the old one.

Full explanation →
130
Multi-Selecthard

Which THREE factors should be considered when designing a Vertex AI Pipeline for continuous training?

Select 3 answers
A.Cost of training and infrastructure
B.Debugging tools like Cloud Debugger
C.Trigger mechanism (time-based or event-based)
D.Number of model versions to keep
E.Data freshness and staleness tolerance
AnswersA, C, E

Budget impacts resource selection.

Why this answer

Cost of training and infrastructure (A) is correct because Vertex AI Pipelines incur compute costs for each pipeline run, including training, data processing, and orchestration. Continuous training amplifies these costs, so you must consider budget constraints, resource optimization (e.g., using preemptible VMs), and cost monitoring to avoid unexpected bills.

Exam trap

Google Cloud often tests the distinction between operational pipeline design factors (triggers, cost, data freshness) and peripheral management tasks (versioning, debugging tools), leading candidates to incorrectly select options like D or B that are valid but not core to pipeline design.

Full explanation →
131
MCQmedium

You need to build a Looker model that joins multiple tables from BigQuery. Which LookML object defines the relationship between tables?

A.JOIN
B.VIEW
C.MODEL
D.EXPLORE
AnswerD

An explore defines the join relationships between views.

Why this answer

In LookML, an EXPLORE defines the starting point and joins to other views. A VIEW defines a single table or derived table. A MODEL contains explores and views.

A JOIN is not a top-level object; joins are defined inside explores.

Full explanation →
132
MCQeasy

A mobile app needs an offline-first NoSQL database that syncs data across devices when connectivity is available. Which Google Cloud database meets these requirements?

A.Memorystore
B.Cloud SQL
C.Bigtable
D.Firestore
AnswerD

Firestore provides offline data persistence and automatic sync, perfect for mobile apps.

Why this answer

Firestore is a NoSQL, serverless, offline-first database that automatically syncs data across devices when connectivity is restored. It provides built-in offline persistence and real-time synchronization, making it ideal for mobile apps that need to work offline and sync later.

Exam trap

The trap here is that candidates may confuse Google Bigtable's NoSQL label with a mobile-friendly NoSQL database, overlooking that Bigtable is designed for high-throughput analytical workloads, not for offline-first mobile sync with real-time listeners.

How to eliminate wrong answers

Option A is wrong because Memorystore is a fully managed in-memory cache (Redis/Memcached) designed for caching and session storage, not a persistent NoSQL database with offline sync capabilities. Option B is wrong because Cloud SQL is a relational (SQL) database, not a NoSQL database, and it does not offer offline-first or cross-device sync features. Option C is wrong because Bigtable is a wide-column NoSQL database optimized for large analytical workloads, not for mobile app offline sync with real-time updates.

Full explanation →
133
MCQeasy

You are loading 10 GB of daily CSV files from a GCS bucket into a BigQuery table. The files contain some malformed rows that you want to skip. Which BigQuery load configuration should you use?

A.Use the 'skip_leading_rows' option.
B.Use the 'ignore_unknown_values' option.
C.Use the 'max_bad_records' option set to a value like 10.
D.Use the 'allow_jagged_rows' option.
AnswerC

max_bad_records specifies the number of allowed bad records; if the number of bad records exceeds this, the load fails.

Why this answer

BigQuery allows setting max_bad_records in load jobs; records exceeding this threshold cause the job to fail. Setting max_bad_records to a value greater than 0 allows the load to succeed while skipping malformed rows.

Full explanation →
134
MCQmedium

A data pipeline processes JSON files from Cloud Storage, transforms them using Apache Beam, and writes the output to BigQuery. Some records are malformed and cause the pipeline to fail. How should the engineer handle these errors to ensure the pipeline continues processing while preserving the malformed records for analysis?

A.Set the pipeline to retry malformed records indefinitely until they succeed.
B.Use a side input to send malformed records to a dead letter queue in Pub/Sub for later reprocessing.
C.Log the malformed records to Stackdriver and skip them in the pipeline.
D.Catch exceptions in a DoFn and write the malformed records to a separate Cloud Storage bucket using a FileIO sink.
AnswerD

This follows the dead letter pattern: malformed records are written to a separate sink, allowing the pipeline to continue and enabling later analysis.

Why this answer

It allows the pipeline to continue processing by catching exceptions within a DoFn and writing malformed records to a separate Cloud Storage bucket using a FileIO sink. This preserves the malformed records for later analysis without blocking the main data flow, which is a standard pattern in Apache Beam for handling dead-letter records. The approach ensures fault tolerance while maintaining data integrity for debugging.

Exam trap

A common misconception in Google Cloud data pipelines is that error handling should involve retrying indefinitely or simply logging errors to Cloud Logging, but the correct approach is to isolate and persist malformed records to a durable sink like Cloud Storage for later analysis.

How to eliminate wrong answers

Option A is wrong because retrying malformed records indefinitely would cause the pipeline to hang or exhaust resources, as malformed records will never succeed due to inherent data issues. Option B is wrong because using a side input to send malformed records to a Pub/Sub dead letter queue is not a direct pattern in Apache Beam; side inputs are for broadcasting data to all elements, not for error handling, and Pub/Sub would require additional setup and does not inherently preserve the records for analysis without a separate sink. Option C is wrong because logging malformed records to Stackdriver and skipping them loses the data permanently, as logs are not designed for structured storage or reprocessing of the original records.

Full explanation →
135
MCQhard

You are migrating an on-premises Kafka cluster to Google Cloud. The cluster has 50 topics with a total throughput of 200 MB/s. You want to minimize operational overhead. Which approach is the most cost-effective?

A.Use Pub/Sub with Kafka-compatible client libraries
B.Use Dataproc to run Kafka on a managed cluster
C.Deploy Kafka on Compute Engine instances with a managed instance group
D.Use Cloud NAT to route traffic to an on-premises Kafka cluster
AnswerB

Why this answer

Managed Kafka services are not available natively on GCP; the recommended approach is to run Kafka on Dataproc, which provides a managed Hadoop/Spark environment with autoscaling.

Full explanation →
136
MCQmedium

A company uses Vertex AI to serve a model. They notice that some predictions are incorrect due to data drift. What is the best way to detect and retrain the model automatically?

A.Store predictions in BigQuery and run scheduled queries
B.Create a Cloud Monitoring dashboard
C.Set up Cloud Logging metrics to monitor predictions
D.Use Vertex AI Model Monitoring with alerts and retraining pipeline
AnswerD

Monitors drift and triggers retraining.

Why this answer

Vertex AI Model Monitoring is specifically designed to detect data drift and feature skew in production models. It can be configured to send alerts and trigger an automated retraining pipeline via Cloud Functions or Vertex AI Pipelines, enabling continuous model improvement without manual intervention. This directly addresses the need for automatic detection and retraining in response to data drift.

Exam trap

The trap here is that candidates may confuse general monitoring tools (Cloud Monitoring, Cloud Logging) with the specialized drift detection and automated retraining capabilities of Vertex AI Model Monitoring, assuming any monitoring solution can trigger retraining without native integration.

How to eliminate wrong answers

Option A is wrong because storing predictions in BigQuery and running scheduled queries is a manual, batch-oriented approach that does not provide real-time drift detection or automated retraining; it requires custom code and lacks native integration with Vertex AI's monitoring capabilities. Option B is wrong because Cloud Monitoring dashboards visualize metrics but do not inherently detect data drift or trigger retraining pipelines; they are for observability, not automated action. Option C is wrong because Cloud Logging metrics can track prediction logs but are not designed for statistical drift analysis (e.g., distribution comparisons) and cannot directly initiate retraining workflows without additional custom logic.

Full explanation →
137
MCQhard

A data analyst runs a complex SQL query in BigQuery that joins multiple large tables and receives the above error. Which action is most likely to resolve the issue?

A.Use a larger number of workers in the query execution.
B.Use smaller tables by sampling data.
C.Add clustering on join columns.
D.Increase the number of slots allocated to the project.
AnswerD

More slots provide more memory and CPU, reducing resource exceeded errors.

Why this answer

The error indicates that the query exceeded the available slot resources in the BigQuery project. Increasing the number of slots allocated to the project (option D) directly addresses this by providing more compute capacity for parallel query execution, which is the correct action to resolve resource exhaustion in BigQuery's serverless architecture.

Exam trap

Google Cloud often tests the misconception that performance tuning (e.g., clustering or sampling) can resolve resource exhaustion errors, when in fact the root cause is insufficient compute capacity that must be addressed by increasing slot allocation.

How to eliminate wrong answers

Option A is wrong because BigQuery automatically manages parallelism; manually specifying a larger number of workers is not supported and would not increase slot capacity. Option B is wrong because sampling data reduces accuracy and may not reflect the full dataset, which is not a valid solution for resource exhaustion—it changes the query result rather than fixing the resource issue. Option C is wrong because clustering on join columns improves query performance and reduces data scanned, but it does not increase the number of slots available; the error is about insufficient compute resources, not about inefficient data access patterns.

Full explanation →
138
Multi-Selectmedium

A data engineer is designing a streaming pipeline with Cloud Pub/Sub and Cloud Dataflow. They need to guarantee at-least-once delivery and handle occasional duplicates. Which TWO configurations should they implement?

Select 2 answers
A.Use idempotent sinks
B.Use global windows with triggers
C.Use fixed windows
D.Use at-least-once Pub/Sub subscription
E.Enable Dataflow Streaming Engine
AnswersA, D

Idempotent sinks allow safe duplicate writes, ensuring exactly-once effect despite duplicates.

Why this answer

Idempotent sinks (e.g., BigQuery with insertId, Cloud Storage with object generation numbers) allow the pipeline to safely process duplicate records without causing data corruption or double-counting. In a streaming pipeline with at-least-once semantics, duplicates are inevitable, and idempotent sinks ensure that repeated writes produce the same result as a single write, maintaining data consistency.

Exam trap

Google Cloud often tests the misconception that windowing strategies (global or fixed) or execution engine features (Streaming Engine) can substitute for explicit delivery guarantees and idempotent sinks, when in fact they address entirely different concerns.

Full explanation →
139
MCQmedium

Your Looker dashboard uses a BigQuery connection. You notice that some queries take over a minute. Which service can you enable to cache results in memory for sub-second Looker queries?

A.BigQuery BI Engine
B.Cloud SQL
C.Cloud Bigtable
D.Cloud Memorystore
AnswerA

BI Engine caches BigQuery data in memory for sub-second queries from BI tools like Looker.

Why this answer

BI Engine is an in-memory analysis service that accelerates BigQuery queries by caching data in memory. It integrates with Looker and Looker Studio. Cloud Memorystore is a Redis/Memcached cache, not directly for BigQuery.

BigQuery BI Engine is the correct service.

Full explanation →
140
Drag & Dropmedium

Drag and drop the steps to create a Cloud Function triggered by Cloud Storage events into the correct order.

Drag steps to the numbered slots on the right, or tap a step then tap a slot.

Steps
Order
1Step 1
2Step 2
3Step 3
4Step 4

Why this order

The correct sequence for creating a Cloud Function triggered by Cloud Storage events is: first ensure the bucket exists (create if needed), then write the function code, deploy the function with the appropriate event trigger (e.g., google.storage.object.finalize), and finally test by uploading a file. Common mistakes include deploying before creating the bucket, testing before deployment, or deploying without code.

Full explanation →
141
Multi-Selectmedium

A data engineer needs to schedule a nightly transfer of data from an Amazon S3 bucket to Cloud Storage. Which two steps are required to achieve this? (Choose TWO.)

Select 2 answers
A.Grant appropriate permissions to the transfer service account
B.Configure a VPC peering between AWS and GCP
C.Use gsutil rsync in a cron job
D.Create a Storage Transfer Service job with a schedule
E.Set up a Cloud Function to copy files
AnswersA, D

Why this answer

The Storage Transfer Service uses a Google-managed service account to access the source S3 bucket. You must grant this service account the appropriate IAM permissions (e.g., `s3:GetObject` and `s3:ListBucket`) on the S3 bucket via an AWS IAM policy. Without these permissions, the transfer job cannot read the data from S3.

Exam trap

Google often tests the misconception that you can use a simple command-line tool like `gsutil rsync` in a cron job for production-scale scheduled transfers, but the correct approach is to use the managed Storage Transfer Service which handles permissions, scheduling, and reliability natively.

Full explanation →
142
MCQmedium

You have deployed a classification model on Vertex AI Endpoints. The model's training data had a balanced class distribution, but over time, the production data has shifted such that one class appears 90% of the time. The model's overall accuracy remains high, but the recall for the minority class has dropped significantly. What is the best approach to detect and address this issue?

A.Retrain the model daily on the entire historical dataset
B.Set up Vertex AI Model Monitoring to detect skew and drift, and retrain using a sliding window of recent data
C.Increase the number of replicas on the endpoint to reduce latency
D.Adjust the decision threshold to improve minority class recall
AnswerB

Model Monitoring detects skew/drift; retraining on recent data adapts to new distribution.

Why this answer

Vertex AI Model Monitoring is specifically designed to detect skew and drift between training and serving data. In this scenario, the production data has shifted to 90% of one class, which is a clear case of data drift. By setting up monitoring, you can be alerted to this drift and then retrain the model using a sliding window of recent data, which adapts to the new distribution without requiring full retraining on the entire historical dataset.

This approach directly addresses the root cause—the shift in class distribution—rather than just treating symptoms.

Exam trap

Google Cloud often tests the distinction between monitoring/detection (Model Monitoring) and reactive fixes (threshold tuning), where candidates mistakenly choose a quick fix like adjusting the decision threshold instead of addressing the root cause of data drift.

How to eliminate wrong answers

Option A is wrong because retraining daily on the entire historical dataset is computationally expensive and does not prioritize recent data; it would still include the old balanced distribution, potentially diluting the model's ability to adapt to the new skewed production data. Option C is wrong because increasing the number of replicas on the endpoint reduces latency and improves throughput, but it does not address data drift or the drop in minority class recall; it is a scaling solution, not a monitoring or retraining solution. Option D is wrong because adjusting the decision threshold can improve recall for the minority class in the short term, but it does not fix the underlying model's inability to generalize to the shifted data distribution; it is a band-aid that may hurt precision and overall model performance.

Full explanation →
143
MCQhard

You manage a BigQuery reservation with 500 baseline slots and autoscaling up to 2000 slots. Your team runs a mix of interactive queries and batch load jobs. During peak hours, you notice that interactive queries are throttled when autoscaling slots are consumed by long-running batch loads. How can you ensure interactive queries get priority access to slots?

A.Create a separate reservation for interactive queries with a higher priority assignment.
B.Reduce the baseline slots to 200 and rely solely on autoscaling.
C.Switch to on-demand pricing to eliminate slot contention.
D.Set the autoscaling max to 1000 slots for batch jobs.
AnswerA

Creating separate reservations for interactive and batch workloads allows you to control slot allocation and prioritize interactive queries.

Why this answer

BigQuery reservations allow you to create separate reservations for different workloads (e.g., interactive queries vs. batch loads) and assign them different priority levels. By creating a dedicated reservation for interactive queries with a higher priority, you ensure that interactive queries get preferential access to slots, even when autoscaling slots are consumed by long-running batch jobs. This directly addresses the contention issue without reducing overall capacity.

Exam trap

Google often tests the misconception that autoscaling alone or reducing baseline slots can solve priority issues, but the key is that without separate reservations and explicit priority assignments, all jobs compete equally for the same pool of slots.

How to eliminate wrong answers

Option B is wrong because reducing baseline slots to 200 and relying solely on autoscaling does not solve the priority issue; autoscaling slots are shared and batch jobs could still consume them, leading to the same throttling of interactive queries. Option C is wrong because switching to on-demand pricing eliminates slot reservations entirely, meaning you lose the ability to guarantee capacity or prioritize workloads, and you may face unpredictable performance and higher costs. Option D is wrong because setting the autoscaling max to 1000 slots for batch jobs does not prevent batch jobs from consuming all available slots; it only limits the maximum they can use, but without priority assignment, interactive queries can still be throttled if batch jobs fill the reservation.

Full explanation →
144
MCQmedium

A company uses GKE to run microservices. They want to ensure the application restarts automatically if it becomes unresponsive. Which probes should they configure in their pod spec?

A.Startup probes only.
B.Liveness probes only.
C.Readiness probes only.
D.Both readiness and liveness probes.
AnswerD

Using both ensures traffic is only sent to ready pods and unresponsive pods are restarted, providing complete health management.

Why this answer

To ensure a pod restarts automatically when it becomes unresponsive, you need a liveness probe. However, the question asks which probes to configure, and the best practice is to use both readiness and liveness probes. The liveness probe restarts the container if it fails, while the readiness probe controls traffic routing, preventing requests from being sent to an unresponsive pod before the liveness probe triggers a restart.

Option D is correct because it combines both for robust health management.

Exam trap

A common misconception is that a liveness probe alone is sufficient for automatic restarts, but the trap here is that readiness probes are also needed to prevent traffic routing to the pod during the restart window, ensuring zero-downtime recovery.

How to eliminate wrong answers

Option A is wrong because startup probes only check if the application has started successfully; they do not monitor ongoing responsiveness or trigger restarts after the initial startup period. Option B is wrong because liveness probes alone will restart the container if unresponsive, but without a readiness probe, traffic may still be routed to the pod during the restart or while it is temporarily unresponsive, causing errors. Option C is wrong because readiness probes only control whether the pod receives traffic; they do not trigger a restart if the application becomes unresponsive, so the pod would remain in a failed state without recovery.

Full explanation →
145
Multi-Selectmedium

You are building a BigQuery table that contains nested and repeated fields (e.g., order with line items). You need to write a query that counts the number of line items per order. Which TWO SQL functions/techniques can you use?

Select 2 answers
A.Window function ROW_NUMBER
B.UNNEST with COUNT
C.STRUCT with aggregation
D.ARRAY_LENGTH
E.SELECT * EXCEPT
AnswersB, D

UNNEST expands the array, then COUNT(*) gives the number of line items per order.

Why this answer

UNNEST flattens the repeated line items array into individual rows, allowing COUNT to aggregate the number of line items per order. Option D is correct because ARRAY_LENGTH directly returns the number of elements in the repeated field array, which corresponds to the line item count.

Exam trap

Google often tests the distinction between functions that operate on arrays directly (like ARRAY_LENGTH) versus those that require row-level expansion (like UNNEST), and candidates may mistakenly choose window functions or STRUCT-based aggregation that do not directly count array elements.

Full explanation →
146
MCQmedium

You run batch predictions using Vertex AI Batch Prediction on a tabular dataset. The job processes 1 million rows and takes 6 hours to complete. You need to reduce the processing time to under 2 hours without increasing cost significantly. What should you do?

A.Switch to a machine type with more CPU cores and vCPUs
B.Increase the machine count (number of worker replicas) in the batch prediction job
C.Downsample the dataset to 500k rows
D.Use Online prediction instead of batch
AnswerB

More workers process data in parallel, reducing runtime linearly with cost.

Why this answer

Vertex AI Batch Prediction supports distributed processing by increasing the number of worker replicas. By adding more workers, the job can process partitions of the 1 million rows in parallel, reducing wall-clock time from 6 hours to under 2 hours. This approach scales horizontally without requiring more expensive machine types, keeping costs roughly linear with the number of workers and avoiding the diminishing returns of vertical scaling.

Exam trap

Google often tests the distinction between vertical scaling (more powerful machines) and horizontal scaling (more machines), where candidates mistakenly choose a more expensive machine type (Option A) instead of the cost-effective parallelization approach (Option B).

How to eliminate wrong answers

Option A is wrong because switching to a machine type with more CPU cores and vCPUs (vertical scaling) typically incurs significantly higher cost per hour, and for batch prediction on tabular data, the bottleneck is often I/O or model inference parallelism, not raw CPU speed; the cost increase would likely exceed the budget constraint. Option C is wrong because downsampling the dataset to 500k rows reduces the amount of data processed, which would lower processing time but also compromises prediction completeness and accuracy, violating the implicit requirement to process all 1 million rows. Option D is wrong because online prediction is designed for low-latency, real-time serving of individual or small batches, not for processing 1 million rows; it would be far more expensive and time-consuming due to per-request overhead and quota limits, and it does not support batch parallelism natively.

Full explanation →
147
MCQhard

Your Dataflow streaming pipeline is experiencing increasing system lag over time. You have enabled autoscaling and the pipeline is using the default streaming engine. Which metric should you monitor in Cloud Monitoring to determine if the pipeline is falling behind due to slow processing or due to a bottleneck in the output sink?

A.Worker CPU utilization
B.System lag
C.Element count
D.Data freshness
AnswerD

Data freshness (time since last output) directly indicates whether the sink is keeping up. High Data freshness suggests a sink bottleneck, while low Data freshness despite high System lag suggests a processing bottleneck.

Why this answer

In Dataflow streaming pipelines, 'System lag' measures the maximum time an item waits to be processed, but it does not by itself distinguish between a processing bottleneck and a sink bottleneck. To differentiate, monitor 'Data freshness' (the time since the last output was written). If Data freshness is high while System lag is also high, the sink is likely the bottleneck.

If System lag is high but Data freshness is low (recent output), the bottleneck is processing. Therefore, the metric that helps determine whether the pipeline is falling behind due to slow processing or a sink bottleneck is Data freshness.

Exam trap

Candidates often assume System lag is the single metric for all delays, but to differentiate between processing and sink bottlenecks, you need to combine System lag with Data freshness.

Full explanation →
148
MCQmedium

Your streaming Dataflow pipeline reads from Pub/Sub, enriches data with a side input, and writes to BigQuery. You need to update the enrichment logic without draining the pipeline, to minimize data loss and maintain exactly-once semantics. What should you do?

A.Cancel the pipeline and create a new one with the updated code.
B.Stop the pipeline, update the code, and restart from the latest snapshot.
C.Use the Dataflow job update mechanism to replace the pipeline with a new version.
D.Drain the pipeline, update the code, and restart with the same job ID.
AnswerC

Dataflow allows updating a streaming pipeline with a new job graph, preserving state and exactly-once processing.

Why this answer

The Dataflow job update mechanism allows you to replace a running pipeline's code with a new version without draining or stopping it, preserving the existing state and minimizing data loss. This mechanism supports exactly-once semantics by ensuring that all in-flight elements are processed exactly once, even after the update, by maintaining the pipeline's checkpoint and watermark state.

Exam trap

The trap here is that candidates often confuse the Dataflow job update mechanism with draining or snapshot-based restarts, not realizing that Dataflow's update feature is specifically designed to allow in-place code changes without data loss or reprocessing.

How to eliminate wrong answers

Option A is wrong because canceling the pipeline would discard all in-flight data and state, leading to data loss and violating exactly-once semantics. Option B is wrong because stopping the pipeline and restarting from a snapshot is not a supported operation in Dataflow; snapshots are used for draining or saving state, but restarting from a snapshot does not guarantee exactly-once processing and can cause data duplication or loss. Option D is wrong because draining the pipeline would allow it to finish processing all existing data before stopping, but then you must create a new pipeline with a new job ID; restarting with the same job ID is not possible after draining, and the drain process itself can cause data loss if not handled correctly.

Full explanation →
149
Multi-Selectmedium

A company needs to store transactional data for a global customer base with strong consistency and 99.999% availability SLA. They anticipate millions of transactions per day across multiple regions. Which TWO storage options meet these requirements? (Choose 2)

Select 2 answers
A.Cloud Bigtable
B.AlloyDB
C.Firestore (in Datastore mode)
D.Cloud SQL (with HA)
E.Cloud Spanner
AnswersC, E

Firestore (in Datastore mode) offers a globally distributed, strongly consistent document database with a 99.999% SLA, meeting both requirements.

Why this answer

Cloud Spanner and Firestore (in Datastore mode) both provide globally distributed, strongly consistent transactions with a 99.999% SLA. Cloud Spanner offers a relational model with ACID across regions, while Firestore (Datastore mode) provides a schemaless document database with strong consistency and multi-region replication. Cloud Bigtable does not offer strong consistency across rows.

AlloyDB and Cloud SQL with HA are regional services, not global, and do not meet the 99.999% SLA across multiple regions.

Exam trap

Candidates may incorrectly assume that only Cloud Spanner meets the strict combination of strong consistency and 99.999% global SLA, but Firestore (Datastore mode) also meets both requirements. Regional services like AlloyDB or Cloud SQL with HA cannot provide global strong consistency across multiple regions.

Full explanation →
150
Multi-Selecteasy

A company is designing a data processing system that must handle both batch and streaming workloads with unified pipeline code. Which two Google Cloud services are most suitable for implementing a unified batch and streaming pipeline? (Choose TWO.)

Select 2 answers
A.Cloud Data Fusion
B.BigQuery
C.Apache Beam SDK
D.Cloud Dataflow
E.Cloud Dataproc
AnswersC, D

Beam is the unified model; Dataflow is one runner.

Why this answer

Apache Beam SDK (C) provides a unified programming model that allows developers to write a single pipeline that can execute in both batch and streaming modes without code changes. It abstracts the underlying execution engine, making it the correct choice for unified pipeline code.

Exam trap

Google Cloud often tests the misconception that Cloud Data Fusion or Cloud Dataproc can achieve unified batch and streaming with a single codebase, but only Apache Beam SDK combined with Cloud Dataflow provides the native programming model and execution engine for this requirement.

Full explanation →
Page 1

Page 2 of 12

Page 3

All pages

123456789101112

Practice PDE by domain

Target a specific domain to shore up weak areas.

Designing Data Processing SystemsIngesting and Processing the DataStoring the DataPreparing and Using Data for AnalysisMaintaining and Automating Data Workloads
See all domains with question counts →