PDE Practice Test 31 — 15 Questions

Question 1

A company uses Cloud Dataflow to process streaming data. They notice that the pipeline's throughput is lower than expected and the system is experiencing high latency. What is the most likely cause?

Accepted Answer

Too few workers. In Cloud Dataflow, streaming pipelines require sufficient worker resources to handle the incoming data rate and maintain low latency. When too few workers are provisioned, the pipeline cannot process data quickly enough, leading to increased backlog and higher latency. This is the most likely cause of reduced throughput and high latency in a streaming pipeline.

Answer

Using batch mode instead of streaming mode

Answer

Too many workers

Answer

Incorrect watermark setting

Question 2

A company needs a fully managed, globally distributed relational database with strong consistency, external consistency, and 99.999% SLA for a financial transaction processing system. Which Google Cloud service should they use?

Accepted Answer

Cloud Spanner. Cloud Spanner is the correct choice because it is a fully managed, globally distributed relational database service that provides strong consistency, external consistency (true serializable transactions across regions), and a 99.999% SLA. These features are essential for a financial transaction processing system that requires ACID compliance and global scalability without sacrificing consistency.

Answer

Firestore

Answer

Bigtable

Answer

Cloud SQL

Question 3

A financial services company uses Vertex AI to serve a fraud detection model. The model was trained on historical data that is updated daily. The team wants to automate retraining when data drift is detected. Which approach best operationalizes this requirement with minimal manual intervention?

Accepted Answer

Enable Vertex AI Model Monitoring for feature drift and skew, then create a Cloud Function that triggers a Vertex AI Pipeline to retrain and deploy the model after validation.. It uses Vertex AI Model Monitoring to automatically detect feature drift or skew, then triggers a Cloud Function that invokes a Vertex AI Pipeline to retrain and redeploy the model after validation. This approach minimizes manual intervention by automating both the detection of data drift and the subsequent retraining and deployment lifecycle.

Answer

Use Cloud Monitoring alerts on prediction latency to trigger a retraining pipeline.

Answer

Manually monitor model performance metrics in Vertex AI Experiments and retrain when accuracy drops.

Answer

Use scheduled Vertex AI Pipelines to retrain the model every night, then deploy automatically.

Question 4

A machine learning pipeline uses Vertex AI Pipelines. One component fails intermittently due to resource constraints. What is the best way to handle this?

Accepted Answer

Use retry policies in the component specification. Vertex AI Pipelines supports retry policies at the component level via the `retry` field in the component specification (YAML or Python). This allows the pipeline to automatically re-execute a failed component when the failure is transient (e.g., resource exhaustion), without manual intervention. Retry policies are the standard mechanism for handling intermittent failures in a serverless orchestration environment like Vertex AI Pipelines.

Answer

Deploy the pipeline on a larger cluster

Answer

Increase the pipeline timeout

Answer

Use a different orchestrator

Question 5

A data engineer needs to perform a one-time migration of 10 TB of data from on-premises Hadoop HDFS to Cloud Storage. The network link is 1 Gbps. Which TWO services or tools should they consider? (Choose 2)

Accepted Answer

Dataproc with DistCp. For a one-time migration of 10 TB from on-premises HDFS to Cloud Storage over a 1 Gbps link, the most appropriate services are Dataproc with DistCp and Transfer Appliance. Dataproc with DistCp is a proven method for copying data from HDFS to Cloud Storage, leveraging Apache Hadoop's distributed copy tool. Transfer Appliance is ideal for large data volumes when network bandwidth is limited, as it physically ships the data. Cloud Storage Transfer Service does not support HDFS as a source directly, so it is not suitable. gsutil rsync with parallel composite uploads could be used but would be slower and less efficient for this volume.

Answer

Cloud Storage Transfer Service

Answer

BigQuery Data Transfer Service

Answer

gsutil rsync with parallel composite uploads

Question 6

A data team is building a near-real-time dashboard that displays aggregated metrics from Kafka topics. They want to use Pub/Sub as a managed messaging service and Dataflow for stream processing. They need to ingest data from Kafka into Pub/Sub with minimal custom code. Which THREE Google Cloud services should they use together? (Choose three.)

Accepted Answer

Dataflow. To ingest data from Kafka into Pub/Sub with minimal custom code, use Kafka Connect with the Pub/Sub connector. Pub/Sub serves as the managed messaging service, while Dataflow provides stream processing for the near-real-time dashboard. Together, these three services meet the requirements without writing extensive custom code.

Answer

Cloud NAT

Answer

Cloud Functions

Question 7

A team needs to optimize online prediction cost for a model that has unpredictable traffic spikes. Which TWO strategies are most effective?

Accepted Answer

Enable autoscaling with a low min_replica_count and high max_replica_count. Autoscaling with a low min_replica_count and high max_replica_count allows the deployment to handle unpredictable traffic spikes by dynamically adjusting the number of replicas. This ensures cost efficiency during low traffic while providing capacity to scale out rapidly when demand surges, a key requirement for online prediction serving.

Answer

Set up Model Monitoring to trigger scaling

Answer

Deploy the model on a single high-memory machine

Answer

Use batch prediction during high traffic

Question 8

A team uses Vertex AI Pipelines to automate retraining of a model every month. The pipeline includes data preprocessing, training, and deployment steps. After a recent update, the pipeline fails intermittently with a timeout error during the deployment step. What is the most likely cause?

Accepted Answer

The trained model size has increased due to more data, causing the deployment step to time out. A model with increased size due to training on more data can cause the deployment step to time out if the deployment infrastructure has a timeout limit. Option A (insufficient permissions) would cause persistent errors, not intermittent. Option C (parallel step execution) typically causes resource contention, not specifically timeout on deployment. Option D (BigQuery quotas) would affect preprocessing, not the deployment step.

Answer

The service account used by the pipeline lacks permissions to deploy the model

Answer

The pipeline is configured to run steps in parallel, leading to resource contention

Answer

BigQuery query quotas are being exceeded during data preprocessing

Question 9

You are configuring Dataplex data quality rules for a BigQuery table. Which three types of rules can be defined using Dataplex's SQL-based rule engine? (Choose three.)

Accepted Answer

Row-level rules (e.g., condition must be true for every row). Dataplex's SQL-based rule engine supports row-level rules that enforce a condition that must be true for every row in a BigQuery table. These rules are defined using a SQL expression that is evaluated per row, and if any row fails the condition, the rule is violated. This allows you to validate data integrity at the most granular level, such as ensuring a column value is always positive.

Answer

Set-level rules (e.g., intersection or difference between two tables)

Answer

Pattern matching rules (e.g., regex on column values)

Question 10

A healthcare organization stores patient data in BigQuery. They need to encrypt a specific column (e.g., SSN) using a key they manage, and decrypt it only for authorized queries via a user-defined function. Which approach should they use?

Accepted Answer

Use BigQuery AEAD encryption functions with a Cloud KMS key. BigQuery AEAD encryption functions (e.g., `AEAD.ENCRYPT` and `AEAD.DECRYPT`) allow you to encrypt a specific column using a customer-managed key stored in Cloud KMS, and then decrypt it only within a user-defined function (UDF) that enforces access controls. This meets the requirement of per-column encryption with key management and authorized decryption via a UDF.

Answer

Use BigQuery column-level access controls

Answer

Use Cloud Key Management Service (Cloud KMS) with CMEK for the BigQuery dataset

Answer

Use Cloud Data Loss Prevention (DLP) to de-identify the column

Question 11

A company wants to orchestrate a multi-step data processing workflow that includes calling a Cloud Run service, waiting for its completion, and then running a BigQuery query. The workflow should be serverless and integrate with Cloud Events. Which Google Cloud service should they use?

Accepted Answer

Cloud Workflows. Cloud Workflows is the correct choice because it is a serverless workflow orchestrator that can coordinate multi-step processes involving Cloud Run and BigQuery. It natively supports waiting for asynchronous operations (like Cloud Run job completion) via its 'call' and 'wait' steps, and it can trigger subsequent steps such as BigQuery queries. Additionally, Cloud Workflows can be triggered by Cloud Events, making it fully integrated with the event-driven architecture described.

Answer

Eventarc

Answer

Cloud Composer

Answer

Cloud Dataflow

Question 12

You need to perform a one-time migration of historical data from an on-premises Teradata data warehouse to BigQuery. The data volume is 50 TB and you have a high-speed network connection (10 Gbps). What is the most efficient way to load the data?

Accepted Answer

Use BigQuery Data Transfer Service for Teradata. BigQuery Data Transfer Service for Teradata is designed for this purpose; it can directly connect to Teradata and transfer data to BigQuery. Exporting to CSV then loading via gsutil is possible but less efficient. Transfer Appliance is for offline transfer but you have high-speed network. Dataproc is not needed.

Answer

Export data from Teradata to CSV files, upload to GCS using gsutil, then load into BigQuery.

Answer

Use Dataproc to run a Spark job that reads from Teradata and writes to BigQuery.

Answer

Use Transfer Appliance to ship the data offline.

Question 13

A data engineer needs to orchestrate a complex data pipeline that involves multiple steps including data extraction from Cloud Storage, transformation using Dataflow, and loading into BigQuery. The pipeline has dependencies between tasks and requires monitoring and retries. Which Google Cloud service should be used for orchestration?

Accepted Answer

Cloud Composer. The question asks for a single service to orchestrate a complex data pipeline with multiple steps, task dependencies, retries, and monitoring. Cloud Composer (a managed Apache Airflow service) is specifically designed for such complex workflow orchestration, supporting dependency graphs, automated retries, and built-in monitoring. While other services (Workflows, Cloud Scheduler, Cloud Tasks) can handle simpler orchestration or event-triggered tasks, they lack the full scheduler, retry, and dependency management capabilities required for this use case. Therefore, Cloud Composer is the correct answer.

Answer

Workflows

Answer

Cloud Scheduler

Answer

Cloud Tasks

Question 14

A data science team uses Vertex AI Pipelines to automate retraining. They want to ensure that only models with performance above a threshold are deployed. Which component should they add to the pipeline?

Accepted Answer

Vertex AI Model Evaluation. Vertex AI Model Evaluation provides built-in evaluation metrics and threshold-based validation that can be used as a pipeline condition to gate model deployment. By adding a Model Evaluation component, the pipeline can compare model performance against a predefined threshold and only proceed to deploy if the metrics (e.g., AUC, precision, recall) meet or exceed the required value.

Answer

Vertex AI Feature Store

Answer

Cloud Build trigger

Answer

Cloud Monitoring alert

Question 15

Which THREE metrics should be monitored for a deployed machine learning model in production?

Accepted Answer

Prediction error rate. Prediction error rate (Option B) is a direct measure of model accuracy in production, reflecting how often the model's predictions deviate from actual outcomes. Monitoring this metric is essential for detecting model degradation, data quality issues, or concept drift that can silently reduce model performance over time.

Answer

Number of replicas

Answer

Training time