PDE Practice Test 28 — 15 Questions

Question 1

A company is designing a data processing pipeline for real-time sensor data. They want to ensure low latency and exactly-once processing semantics. Which two Google services should they combine to achieve this? (Choose 2)

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

Accepted Answer

Cloud Pub/Sub with exactly-once delivery. Cloud Pub/Sub with exactly-once delivery (Option C) ensures that each message is delivered to subscribers exactly once, preventing duplicates in the pipeline. Cloud Dataflow with exactly-once processing mode (Option D) provides end-to-end exactly-once semantics by leveraging consistent snapshots and idempotent sinks, which is critical for real-time sensor data pipelines requiring low latency and accuracy.

Answer

Cloud Dataproc with Spark Streaming

Answer

Cloud Functions with Cloud Pub/Sub triggers

Answer

Cloud IoT Core with device gateways

Question 2

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

Accepted Answer

AlloyDB. 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.

Answer

Cloud Spanner

Answer

Cloud SQL for PostgreSQL

Answer

BigQuery

Question 3

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?

Accepted Answer

Increase v2's traffic split by 10% each day while also adding replicas for v2 based on CPU utilization.. The best approach is to gradually increase v2's traffic while dynamically scaling replicas based on CPU utilization. This allows controlled rollout, monitoring of latency, and ensures sufficient resources to maintain the latency SLO. Adding replicas based on CPU utilization helps handle increased load as traffic to v2 grows, preventing latency degradation.

Answer

Merge v2's model into v1 by retraining v1 with v2's architecture and deploy as a single model.

Answer

Immediately set v2 to serve 100% traffic and monitor latency; if it exceeds 200 ms, roll back.

Answer

Use a separate endpoint for v2 and route traffic at the load balancer level.

Question 4

Your company is building a data processing system that ingests sensor data from millions of devices, processes it in near real-time to detect anomalies, and stores raw and processed data for long-term analytics. The system must meet a 99.9% uptime SLA and minimize data loss. Which THREE design choices are best? (Choose three.)

Accepted Answer

Use Cloud Pub/Sub as the ingestion layer with a dead-letter topic to capture unprocessed messages.. Option A is correct because Cloud Pub/Sub provides a highly available, globally distributed ingestion layer that decouples millions of devices from downstream processing, and a dead-letter topic captures messages that repeatedly fail processing so they are not silently lost, supporting the 99.9% uptime and minimal data-loss goals. Option C is correct because Dataflow with at-least-once processing guarantees ensures no message is dropped during failures, and performing deduplication downstream handles the duplicate records that at-least-once semantics can produce, which is the standard trade-off for minimizing data loss in near real-time pipelines. Option D is correct because Cloud Storage is the durable, low-cost archival store for raw sensor data, while BigQuery is the serverless analytics warehouse suited for long-term processed-data analytics at scale. Option B is not best because Bigtable is optimized for high-throughput random reads/writes of time-series or keyed data rather than long-term analytical querying, and Cloud Storage alone for processed data lacks BigQuery's SQL analytics capability. Option E is not appropriate because a global Cloud Load Balancer distributes HTTP(S)/TCP traffic to backends and is not used to front Dataflow workers, which are managed by the Dataflow service and scale internally.

Answer

Store raw data in Cloud Bigtable and processed data in Cloud Storage.

Answer

Use a global Cloud Load Balancer in front of the Dataflow workers.

Question 5

Your team is processing a large dataset with Apache Beam on Dataflow. The pipeline sometimes fails due to transient errors when writing to a BigQuery sink. You need to ensure that failed records are not lost and can be reprocessed later without blocking the pipeline. What is the best approach?

Accepted Answer

Use a side output (e.g., via TupleTag) to write failed records to a dead letter sink (e.g., GCS or Pub/Sub) and continue processing the main output.. Using a dead letter pattern with a side output to write failed records to a GCS bucket (or Pub/Sub) allows the pipeline to continue processing healthy records while failed records are stored for later analysis and reprocessing.

Answer

Configure the pipeline to use at-least-once semantics and rely on Dataflow to retry the entire bundle.

Answer

Increase the number of workers to reduce the chance of transient errors.

Answer

Use a try-catch block in the DoFn and log the error; continue processing other elements.

Question 6

A data team wants to use Approximate Aggregation Functions in BigQuery to get faster query results. Which two functions can they use? (Choose 2)

Accepted Answer

APPROX_QUANTILES. APPROX_QUANTILES [CORRECT] is a valid BigQuery approximate aggregation function that returns approximate quantile boundaries for a dataset, letting the team trade a small amount of accuracy for much faster results on large data. APPROX_COUNT_DISTINCT [CORRECT] is also a real BigQuery approximate aggregation function that estimates the number of distinct values using HyperLogLog++ sketches, providing fast, scalable cardinality estimates. The other options are not BigQuery functions: APPROX_SUM, APPROX_AVG, and APPROX_MEDIAN do not exist in BigQuery's function library, so they cannot be used for approximate aggregation.

Answer

APPROX_SUM

Answer

APPROX_AVG

Answer

APPROX_MEDIAN

Question 7

A data engineer is building a Dataflow pipeline in Python that reads from BigQuery, transforms data, and writes to Cloud Storage. The pipeline will be deployed in production. Which approach should they use to ensure the pipeline is reusable across environments with different configuration parameters?

Accepted Answer

Use a Dataflow Flex Template. Dataflow Flex Templates allow you to package a Docker container with your pipeline code and dependencies, enabling parameterization at runtime via the Dataflow UI, CLI, or API. This makes the pipeline reusable across environments (e.g., dev, staging, prod) by passing different configuration parameters (like project IDs, table names, or output paths) without modifying the code. Flex Templates support custom container images and are the recommended approach for production pipelines that need environment-agnostic deployment.

Answer

Create a separate pipeline for each environment with hardcoded values

Answer

Use a Dataflow Classic Template

Answer

Run the pipeline using the DirectRunner for each environment

Question 8

A data engineer wants to set up automatic deletion of objects from a Cloud Storage bucket after 30 days, and transition objects older than 7 days to Nearline storage. Which THREE steps should they take? (Select three.)

Accepted Answer

Create a lifecycle rule with action Delete and condition age 30 days. Option B is correct because a lifecycle rule with the Delete action and an age condition of 30 days will automatically remove objects once they reach 30 days old, satisfying the deletion requirement. Option C is correct because a lifecycle rule with the SetStorageClass action targeting Nearline and an age condition of 7 days transitions objects older than 7 days to Nearline storage, exactly matching the transition requirement. Option E is correct because lifecycle rules must be applied to the bucket by submitting the JSON configuration with the gsutil lifecycle set command (or equivalent API call); creating rules without applying them has no effect. Option A is not needed because changing the bucket's default storage class only affects newly uploaded objects and does not perform age-based transitions or deletions. Option D is not required because Object Versioning is unrelated to time-based deletion or storage-class transitions and would actually complicate deletion by retaining noncurrent versions.

Answer

Set the bucket's default storage class to Nearline

Answer

Enable Object Versioning on the bucket

Question 9

You are using BigQuery ML to train a matrix factorization model for a recommendation system. The training data consists of user-item interactions. You notice that the model is overfitting. Which of the following hyperparameter changes would most likely reduce overfitting?

Accepted Answer

Increase w_reg (regularization weight) from 0.1 to 0.5. Increasing w_reg raises the L2 regularization penalty applied to the learned latent factor matrices, which shrinks factor magnitudes and constrains model complexity. In BigQuery ML's matrix factorization, w_reg directly controls how strongly the optimizer penalizes large weights, so moving from 0.1 to 0.5 tightens the fit and reduces overfitting on sparse user-item interaction data.

Answer

Decrease w_reg (regularization weight) from 0.1 to 0.01

Answer

Increase num_factors from 10 to 20

Answer

Increase num_training_iterations from 10 to 20

Question 10

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?

Accepted Answer

Cloud Spanner. 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.

Answer

Cloud SQL with cross-region replicas

Answer

Firestore

Answer

Cloud Bigtable

Question 11

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?

Accepted Answer

Create a reservation in the Administration panel and assign it to the project.. 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.

Answer

Enable BI Engine by setting the dataset option 'enable_bi_engine=TRUE' in the dataset metadata.

Answer

Grant all Looker Studio users the 'biengine.user' IAM role on the project.

Answer

Create materialized views of the tables and connect Looker Studio to the views.

Question 12

A data engineer needs to monitor the performance of BigQuery queries to identify opportunities for optimization. Which TWO metrics should they focus on? (Choose two.)

Accepted Answer

Slot usage. Slot usage (A) is correct because BigQuery charges and throttles based on slots, and monitoring slot utilization reveals whether queries are CPU-bound, queuing, or over-consuming resources, which directly points to optimization opportunities such as reducing complexity or using reservations. Data scanned per query (B) is correct because BigQuery bills on bytes processed, and tracking scanned data highlights inefficient patterns like SELECT *, missing partition filters, or unclustered tables that can be optimized to cut cost and improve speed. Query execution time (C) is not one of the two targeted metrics here since it is an outcome rather than a direct optimization lever, and the question asks for metrics that reveal optimization opportunities. Number of tables joined (D) is not a standard BigQuery performance metric and does not by itself indicate inefficiency. Number of users (E) is a usage/access statistic, not a query performance metric relevant to optimization.

Answer

Query execution time

Answer

Number of tables joined

Answer

Number of users

Question 13

A company processes real-time clickstream data from websites. They need to aggregate user sessions that may span multiple hours and handle events that arrive late due to network delays. The pipeline must avoid discarding late data. Which Dataflow feature should they configure?

Accepted Answer

Use session windows with a gap duration and allow late data with a suitable allowed_lateness. Session windows are ideal for aggregating user sessions that span multiple hours, as they group events based on a gap duration of inactivity. By configuring `allowed_lateness`, the pipeline can handle late-arriving events without discarding them, ensuring completeness. This directly addresses the requirement to avoid discarding late data while aggregating sessions.

Answer

Use fixed windows with a trigger that fires after every element

Answer

Use the GlobalWindow with a watermark

Answer

Use sliding windows with no allowed lateness

Question 14

A company wants to implement model monitoring for a deployed classification model. Which three types of monitoring should they set up? (Select 3)

Accepted Answer

Training-serving skew. For a deployed classification model, the three essential monitoring types are training-serving skew (B), prediction drift (C), and input feature drift (D). Training-serving skew (B) is correct because it detects discrepancies between the feature transformations applied during training and those applied at inference time, which can silently degrade model accuracy in production. Prediction drift (C) is correct because it tracks changes in the distribution of the model's output predictions over time, signaling that the model's behavior is shifting even when ground truth labels are unavailable. Input feature drift (D) is correct because it monitors changes in the distribution of incoming feature values relative to the training data, which is often an early warning that the model is operating on data unlike what it learned from. Infrastructure cost monitoring (A) is not a model-quality monitoring type and does not detect model degradation, and model version comparison (E) is a deployment/management activity rather than an ongoing monitoring category for a deployed classifier.

Answer

Infrastructure cost monitoring

Answer

Model version comparison

Question 15

A company uses Pub/Sub to ingest events from multiple sources. They need to ensure that messages from a specific source are processed in order (per source partition). They also need to deduplicate messages. Which TWO features should they use?

Accepted Answer

Enable exactly-once delivery on the subscription. Option E is correct because Pub/Sub message ordering is enabled by setting an ordering key on messages; messages sharing the same ordering key are delivered to subscribers in the order they were published, which satisfies the per-source-partition ordering requirement. Option C is correct because enabling exactly-once delivery on the subscription ensures that messages are not redelivered after successful acknowledgment, providing the required deduplication behavior. Option A is incorrect because schemas validate message format and structure, not ordering. Option B is incorrect because a dead letter topic only captures messages that cannot be successfully processed after retries, and does not reorder or deduplicate messages. Option D is incorrect because a large ack deadline only extends the processing window to reduce premature redelivery; it does not guarantee ordering or exactly-once deduplication.

Answer

Set a message schema to enforce ordering

Answer

Use a dead letter topic to handle out-of-order messages

Answer

Use a pull subscription with a large ack deadline