PDE Practice Test 38 — 15 Questions

Question 1

A data engineer needs to create a unified table that combines data from Cloud Storage (Parquet files) and BigQuery native tables, with fine-grained access control and governance. Which three Google Cloud features should they use together? (Choose THREE.)

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

BigQuery. BigQuery is correct because it serves as the unified query engine that can read data from both Cloud Storage (via external tables or BigLake) and native BigQuery tables, enabling a single SQL interface for analysis. It also integrates with fine-grained access control through row-level security and column-level access policies, and supports governance via Data Catalog and VPC Service Controls.

Answer

Dataproc

Answer

Cloud SQL

Question 2

A company is designing a streaming data pipeline to process real-time clickstream events. They need to aggregate events by session window with a 5-minute gap and enable exactly-once processing semantics. Which Google Cloud service should they use?

Accepted Answer

Cloud Dataflow with Apache Beam. Cloud Dataflow with Apache Beam is the correct choice because it provides native support for session windows with a 5-minute gap duration and exactly-once processing semantics via its sink and source integrations. Dataflow's Beam SDK allows you to define session windows using `Window.into(Sessions.withGapDuration(Duration.standardMinutes(5)))`, and its checkpointing and idempotent writes ensure exactly-once delivery even in failure scenarios.

Answer

Cloud Pub/Sub with Cloud Functions

Answer

Cloud Dataproc with Spark Streaming

Answer

Cloud Bigtable with Dataflow templates

Question 3

Which TWO metrics are most important to monitor for a real-time online prediction system to ensure service reliability and model performance?

Accepted Answer

Prediction latency (p50, p99). Prediction latency (p50, p99) is critical because it directly impacts user experience and system reliability; high tail latency (p99) can indicate resource contention or model complexity issues. Prediction error rate (4xx/5xx) is essential for detecting serving infrastructure failures, such as model server crashes or misconfigured endpoints, which degrade service reliability. Both metrics provide real-time visibility into the serving layer's health and performance, distinct from offline training metrics.

Answer

Feature distribution skew between training and serving

Answer

Number of training examples used for the latest model version

Answer

Batch prediction job throughput

Question 4

A data science team uses Cloud Build and Vertex AI to implement CI/CD for their machine learning models. Which THREE steps are essential for a production-ready operationalization pipeline? (Choose 3.)

Accepted Answer

Deploy the model to a staging endpoint for manual approval before promoting to production.. Deploying to a staging endpoint for manual approval before promoting to production is a critical step in a production-ready CI/CD pipeline. This allows data scientists to validate model behavior, performance, and fairness in a near-production environment, preventing regressions and ensuring governance compliance before the model serves live traffic.

Answer

Store all training artifacts in Cloud Storage without versioning.

Answer

Automatically deploy every new model version directly to the production endpoint.

Question 5

A data pipeline processes streaming data from Pub/Sub to BigQuery. The pipeline needs to handle late-arriving data that is up to 1 hour late. Which Dataflow feature should be used?

Accepted Answer

Fixed windows with allowed lateness. Fixed windows with allowed lateness are the correct choice because the pipeline needs to handle late-arriving data up to 1 hour late while processing data in fixed time intervals (e.g., 1-hour windows). The `allowedLateness` parameter in Dataflow (Apache Beam) allows late data to be included in the appropriate fixed window for up to the specified duration after the watermark passes the window end. This ensures that late Pub/Sub messages are correctly joined with their original window in BigQuery.

Answer

Global windows with watermark

Answer

Session windows

Answer

Sliding windows with allowed lateness

Question 6

You need to estimate the cost of a BigQuery query before running it. Which command or feature should you use?

Accepted Answer

Use the bq command with the --dry_run flag.. The `bq` command with the `--dry_run` flag allows you to estimate the amount of data a BigQuery query will process before actually executing it. This dry run does not read any data or incur charges; it simply returns the estimated bytes to be processed, which you can use to calculate the cost based on BigQuery's pricing model.

Answer

Check the BigQuery jobs list for similar queries.

Answer

Use the BigQuery cache to estimate if the query is cached.

Answer

Run EXPLAIN on the query to see the query plan.

Question 7

A company wants to ingest IoT sensor data from thousands of devices into BigQuery for near-real-time analytics. The data volume is approximately 10 GB per hour. Which combination of Google Cloud services should they use for a cost-effective and scalable solution?

Accepted Answer

Pub/Sub → Dataflow → BigQuery. Pub/Sub provides a scalable, managed ingestion layer for high-volume IoT data, decoupling producers from consumers. Dataflow (Apache Beam) processes the streaming data in near-real-time with exactly-once semantics and auto-scaling, writing directly to BigQuery for analytics. This combination minimizes operational overhead and cost by avoiding intermediate storage and manual scaling.

Answer

Cloud IoT Core → Cloud Functions → BigQuery

Answer

Cloud IoT Core → Cloud Dataproc → BigQuery

Answer

Cloud IoT Core → Cloud Storage → BigQuery load jobs

Question 8

A company uses Cloud Dataflow to process streaming data from Pub/Sub into BigQuery. The pipeline uses a side input from a Cloud Bigtable table containing user profile information to enrich the events. The side input is updated every hour. Which approach should the company use to ensure that the pipeline uses the latest profile data without causing high memory usage?

Accepted Answer

Use a side input that is periodically refreshed by reading the Cloud Bigtable table at a regular interval.. Cloud Dataflow supports periodically refreshing side inputs by reading from an external source like Cloud Bigtable at a specified interval. This approach keeps the profile data up-to-date without storing the entire side input in memory for the lifetime of the pipeline; instead, the side input is rebuilt and cached only when refreshed, controlling memory usage.

Answer

For each incoming event, read the corresponding profile from Cloud Bigtable using a synchronous call.

Answer

Use a CoGroupByKey transform to join the stream with a bounded PCollection created from the Cloud Bigtable table.

Answer

Stream the profile updates into a separate BigQuery table and use a BigQuery streaming query to join in real-time.

Question 9

Which TWO configurations are required to enable online prediction for a model deployed on Vertex AI Endpoints?

Accepted Answer

The endpoint must be configured with a machine type (e.g., n1-standard-2).. Vertex AI Endpoints require a machine type to be specified when deploying a model. The machine type determines the compute resources (CPU/memory) allocated to the serving container, which is essential for handling prediction requests. Without a machine type, the endpoint cannot provision the underlying infrastructure to serve online predictions.

Answer

A feature store must be attached to the endpoint.

Answer

The model must be trained on Vertex AI.

Answer

Autoscaling must be enabled.

Question 10

A company is migrating on-premises Hadoop Hive workloads to Google Cloud. They want to use Dataproc for Spark processing and require a managed Hive metastore that can be shared across multiple Dataproc clusters. Which TWO components should they use?

Accepted Answer

Dataproc Metastore. The question specifies a managed Hive metastore that can be shared across multiple Dataproc clusters. Dataproc Metastore is a fully managed service that provides a Hive metastore, requiring no manual setup. Cloud SQL for MySQL can serve as a metastore backend, but it is not a managed service—it requires manual configuration, maintenance, and is not a dedicated metastore service. Therefore, only Dataproc Metastore meets the criteria of a managed Hive metastore.

Answer

Cloud Bigtable

Answer

Dataproc on GKE

Answer

Dataproc Serverless

Answer

Cloud SQL for MySQL

Question 11

Refer to the exhibit. What is the most likely cause?

Accepted Answer

The model container does not support this prediction route. The error indicates that the model container does not have a route configured to handle the specific prediction request. In Vertex AI or similar MLOps platforms, each model container must expose a prediction endpoint (e.g., /predict or /v1/models/{model}:predict) via a route defined in the serving configuration. If the container only supports batch prediction (e.g., via a custom gRPC service) or lacks the required HTTP route, the request will fail with this error.

Answer

The request format is incorrect

Answer

The model was built for batch prediction only

Answer

The endpoint ID is wrong

Question 12

A team is deploying a complex model with multiple preprocessing steps. They want to ensure consistent preprocessing during training and serving. Which three approaches can achieve this? (Select 3)

Accepted Answer

Store preprocessing logic in a shared Python module. Storing preprocessing logic in a shared Python module ensures that the same code is used during both training and serving, eliminating drift between environments. This approach leverages version control and dependency management to guarantee consistency, which is critical for reproducibility in production ML pipelines.

Answer

Use a separate preprocessing service called from the model

Answer

Use two separate pipelines for training and serving

Question 13

Which Dataflow feature automatically scales the number of workers based on the pipeline's current workload, and also selects the optimal machine type for each worker based on the pipeline's resource requirements?

Accepted Answer

Dataflow Prime. Dataflow Prime is the correct answer because it is the only Dataflow feature that provides both automatic worker scaling (horizontal autoscaling) and intelligent machine type selection (vertical autoscaling). It dynamically adjusts the number of workers based on the pipeline's current workload and selects the optimal machine type (e.g., CPU, memory, or accelerator-optimized) for each worker based on the pipeline's resource requirements, such as CPU utilization, memory pressure, or shuffle throughput.

Answer

Dataflow Shuffle

Answer

Dataflow Streaming Engine

Answer

Dataflow Flex Templates

Question 14

A company is migrating their on-premises Apache Spark jobs to Google Cloud Dataproc. They want to minimize operational overhead and cost for jobs that run only a few times per day. Which TWO strategies should they adopt? (Choose TWO.)

Accepted Answer

Store all data in Cloud Storage instead of HDFS, and use the Cloud Storage connector to access it.. Storing data in Cloud Storage decouples storage from compute, allowing ephemeral clusters to be spun up and down without data loss. The Cloud Storage connector provides Hadoop-compatible file system access, eliminating the need for HDFS replication and reducing costs by avoiding persistent cluster storage. Option D is correct because ephemeral Dataproc clusters are created per job and deleted after completion, which minimizes cost and operational overhead for intermittent workloads, as there is no need to maintain a persistent cluster. Options A and B are incorrect: A proposes HDFS replication, which is unnecessary when using Cloud Storage, and B suggests rewriting jobs as Dataflow pipelines, which is not a required strategy for the stated goal of minimizing overhead for existing Spark jobs. Option E is incorrect because a persistent cluster incurs continuous costs and operational overhead, which is not optimal for jobs that run only a few times per day.

Answer

Configure HDFS replication factor to 3 to ensure data durability during cluster restarts.

Answer

Rewrite the Spark jobs as Dataflow pipelines to take advantage of serverless processing.

Answer

Use a small persistent cluster that runs continuously and submit jobs to it.

Question 15

A data engineering team needs to build a data integration pipeline that involves connecting to multiple sources, performing data transformations with visual editing, and then running custom machine learning algorithms. The team has both data analysts and data scientists. Which approach is most suitable?

Accepted Answer

Use Cloud Data Fusion for the initial ingestion and transformations, then export the data to Cloud Dataproc for the ML algorithms. It leverages Cloud Data Fusion's visual, no-code interface for data ingestion and transformation, which is ideal for data analysts, and then exports the prepared data to Cloud Dataproc, which provides native support for custom machine learning algorithms using Spark or Hadoop, meeting the data scientists' needs. This separation of concerns optimizes the pipeline for both user groups and avoids forcing all tasks into a single tool that may not excel at both visual ETL and custom ML.

Answer

Use Cloud Composer to orchestrate both Data Fusion and Dataproc

Answer

Use only Cloud Dataproc for all steps

Answer

Use only Cloud Data Fusion for all steps