How many Refer to the Exhibit Practice Questions questions are on this page?

This page has 12 Refer to the Exhibit Practice Questions scenario questions for the PDE exam, each with detailed explanations and wrong-answer analysis.

How should I approach PDE scenario questions?

Read the full scenario before looking at the answer options. Identify the constraint or requirement in the scenario, then eliminate options that are generally true but wrong for this specific case. Scenario questions reward careful reading over pattern matching.

← Back to Google Professional Data Engineer questions

Scenario-based practice

Refer to the Exhibit Practice Questions

Practise Google Professional Data Engineer practice questions — original exam-style scenarios covering every exam domain, with detailed explanations, wrong-answer analysis, and common exam traps.

Start full practice test Read exam guide

scenario questions

PDE

exam code

Google Cloud

vendor

Scenario guide

How to approach refer to the exhibit practice questions

Practise exhibit-style questions that ask you to read a topology, table, command output or diagram before choosing the best answer.

Quick answer

Exhibit-style questions test whether you can read a topology, command output, diagram or table before choosing the best answer.

How to extract the relevant detail from an exhibit.

How topology, command output or routing information affects the answer.

How to avoid answering from memory before reading the evidence.

How to map the exhibit back to the exam objective.

Practice scenarios

Question 1hardmultiple choice

Full question →

Refer to the exhibit. A team is trying to run a custom prediction container on Vertex AI Endpoint. They get this error when the container starts. What is the most likely cause?

Exhibit

Log: "Container failed with error: exec format error. Ensure the container has an entry point."

A
The container image is too large
Why wrong: Size does not cause exec format error.
B
The entry point is missing or incorrect
The error message directly states to ensure the container has an entry point.
C
The container is built for a different CPU architecture
Why wrong: Would cause a different error like 'exec format error' possible but less specific; the error message explicitly mentions entry point.
D
The model file is missing from the container
Why wrong: Would cause a runtime error, not exec format.

Full breakdown with real-world context →

Question 2mediummultiple choice

Full question →

A BigQuery query fails with the error shown in the exhibit. What is the most likely cause?

Exhibit

Refer to the exhibit.

Exhibit:
Error: Resources exceeded during query execution.
Query statement: SELECT * 
FROM `project.dataset.table` 
WHERE date >= '2023-01-01'

A
The query scans too many partitions or data without efficient pruning
SELECT * on a large table can exceed resource limits; partition pruning might help.
B
The table has too many partitions
Why wrong: BigQuery handles thousands of partitions without issue.
C
The user does not have permission to query the table
Why wrong: Permission errors would be access denied, not resource exceeded.
D
Insufficient slot capacity in the project
Why wrong: Slot capacity issues result in a different error about unavailable resources.

Full breakdown with real-world context →

Question 3mediummultiple choice

Full question →

Refer to the exhibit. A data scientist deploys a model using this configuration. Users report that after a few hours of inactivity, the first prediction request takes over 30 seconds. What is the most likely cause?

Network Topology

A
The automatic scaling configuration allows scaling down to zero replicas, causing a cold start on the first request.
minReplicaCount: 0 permits scaling to zero, and after inactivity, the first request must wait for a new replica to start.
B
The network latency between the client and the endpoint is high due to regional distance.
Why wrong: Network latency would affect all requests, not just the first after inactivity.
C
The endpoint is misconfigured with the wrong regional endpoint.
Why wrong: The endpoint URL looks correct for us-central1; misconfiguration would cause persistent failures.
D
The model is too large and exceeds the instance memory.
Why wrong: A memory issue would likely cause errors, not just slow first request.

Full breakdown with real-world context →

Question 4hardmultiple choice

Full question →

Based on the exhibit, what is the most likely cause of duplicate rows despite using the same event_id as insertId?

Exhibit

Refer to the exhibit.

```
# BigQuery table schema and sample data
Table: mydataset.events
Columns:
  event_id: STRING (REQUIRED)
  event_timestamp: TIMESTAMP (REQUIRED)
  event_data: STRING (NULLABLE)
  user_id: STRING (REQUIRED)
Partitioned by: event_timestamp (daily)
Clustered by: user_id

Job: Dataflow pipeline writing 1000 events/second to this table using streaming inserts with insertId = event_id.

Monitoring shows intermittent 'duplicate rows' in queries that count distinct event_ids.
```

A
BigQuery's streaming buffer deduplication is best-effort and may not catch duplicates within a short time window.
Duplicate inserts within milliseconds can bypass dedup due to coarseness.
B
The Dataflow pipeline is retrying inserts due to network errors, and the same event_id is not being used in retries.
Why wrong: If the insertId is the same, retries should be deduplicated.
C
The pipeline is writing more than 100,000 rows per second, exceeding BigQuery's streaming quota.
Why wrong: 1000 rows/second is well within quota (100,000 rows/second).
D
The table is partitioned by timestamp, so BigQuery cannot deduplicate across partitions.
Why wrong: Partitioning does not affect deduplication logic based on insertId.

Full breakdown with real-world context →

Question 5hardmultiple choice

Full question →

A Dataflow pipeline as described in the exhibit has increasing lag. Which optimization is most likely to reduce the lag?

Exhibit

Refer to the exhibit.

Exhibit:
Pipeline description:
- Source: PubSubIO.read()
- Transform: ParDo(Process)
- Window: Window.into(FixedWindows of 1 minute)
- Transform: GroupByKey
- Sink: Write to BigQuery using StreamingInserts
- Estimated throughput: 10MB/s
- Observed lag: increasing

A
Use FileLoads instead of StreamingInserts for BigQuery output
FileLoads (batch loads) are more efficient for high throughput and reduce lag.
B
Increase the number of workers
Why wrong: More workers may help but the streaming insert bottleneck remains.
C
Use global windows instead of fixed windows
Why wrong: Global windows would increase latency as data waits until end.
D
Add additional ParDo transforms
Why wrong: More transforms add processing overhead, increasing lag.

Full breakdown with real-world context →

Question 6hardmultiple choice

Full question →

A company runs a Cloud Dataflow streaming pipeline that reads from Cloud Pub/Sub, performs a fixed window of 10 seconds, joins with a slowly-changing dimension table stored in Cloud Bigtable, and writes results to BigQuery. The pipeline has been running for months but recently started exhibiting increasing latency and occasional data loss. The pipeline uses default settings with autoscaling enabled (min 2, max 20 workers). The Bigtable cluster has 3 nodes. The dimensions are updated infrequently. The latency has grown from seconds to minutes. Examining the Dataflow monitoring UI, you see that the 'System Lag' metric is increasing, and some windows are not being emitted. The CPU utilization on Bigtable nodes is below 50%. There are no errors in the logs. Which action is most likely to resolve the issue?

A
Set the pipeline option --maxNumWorkers to a value between 5 and 10.
Prevents over-scaling and shuffle overhead.
B
Increase the window duration to 30 seconds to reduce the number of windows.
Why wrong: Doesn't address system lag.
C
Redesign the pipeline to use a side input for the dimension table instead of a lookup.
Why wrong: Side input may cause similar issues.
D
Increase the number of Bigtable nodes to reduce lookup latency.
Why wrong: Bigtable CPU is low, so not the bottleneck.

Full breakdown with real-world context →

Question 7mediummultiple choice

Full question →

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

Exhibit

Error: Vertex AI.Exception: 400 Failed to deploy model to endpoint projects/.../endpoints/1234. Details: The resource 'projects/.../models/5678' is missing an artifact URI. Please upload the model artifact to Cloud Storage and create a new model version.

A
The model artifact was not uploaded to Cloud Storage
The error explicitly states the artifact URI is missing.
B
The endpoint does not exist
Why wrong: The error indicates the endpoint exists but the model artifact is missing.
C
The service account lacks permissions
Why wrong: Permissions would cause a different error, not 'missing artifact URI'.
D
The model ID is invalid
Why wrong: The model ID is referenced correctly; the issue is the missing artifact URI.

Full breakdown with real-world context →

Question 8mediummultiple choice

Full question →

Refer to the exhibit. A developer sees this log entry when trying to get a prediction. What is the most likely cause?

Exhibit

{
 "severity": "ERROR",
 "message": "Prediction failed: Model 'projects/my-project/models/12345/versions/v1' not found.",
 "timestamp": "2024-01-20T10:00:00Z",
 "request": "POST /v1/projects/my-project/models/12345:predict"
}

A
The model ID is incorrect
Why wrong: The model ID '12345' might be correct; the error specifically says version not found, not model not found.
B
The model version is not deployed
A model version must be deployed to an endpoint to serve predictions; 'not found' suggests it is not deployed.
C
The endpoint does not exist
Why wrong: The request is to the model resource, not an endpoint; the error is about the model version.
D
The project ID is wrong
Why wrong: The project ID is included in the model path; if incorrect, the error would be different.

Full breakdown with real-world context →

Question 9easymultiple choice

Full question →

Refer to the exhibit. A Cloud Build step fails when pushing a Docker image to Artifact Registry. What is the missing IAM role for the Cloud Build service account?

Exhibit

Refer to the exhibit.
```json
{
  "error": "denied: permission denied for us-central1-docker.pkg.dev/my-project/my-repo/my-model:latest"
}
```

A
roles/artifactregistry.writer
This role allows pushing images to Artifact Registry.
B
roles/containerregistry.admin
Why wrong: The error is from Artifact Registry, not Container Registry.
C
roles/storage.objectCreator
Why wrong: ObjectCreator is for Cloud Storage, not Artifact Registry.
D
roles/cloudbuild.builds.editor
Why wrong: This role allows creating builds but does not grant push permission to Artifact Registry.

Full breakdown with real-world context →

Question 10hardmultiple choice

Full question →

Refer to the exhibit. A data engineer sees these metrics from Cloud Monitoring for a deployed Vertex AI Endpoint. What is the most effective action to reduce latency?

Exhibit

Metric: CPU Utilization (Model Endpoint)
Current: 90%
Threshold: 80%
Trend: Increasing

Metric: Prediction Latency (p95)
Current: 1500ms
Threshold: 1000ms
Trend: Increasing

A
Switch to batch prediction
Why wrong: Batch prediction is not for reducing online latency; it is for asynchronous processing.
B
Increase the number of replicas
Adding replicas scales horizontally, reducing load per replica and improving latency.
C
Reduce the machine type
Why wrong: Reducing machine type would likely worsen the issue.
D
Enable model quantization
Why wrong: Quantization can reduce model size and improve inference speed, but the immediate problem is high CPU utilization due to traffic; scaling is more direct.

Full breakdown with real-world context →

Question 11hardmultiple choice

Full question →

Refer to the exhibit. The feature store 'my_fs' responds to offline queries but online serving requests fail. What is the most likely cause?

Exhibit

gcloud ai featurestores describe projects/.../locations/us-central1/featurestores/my_fs
Output:
online_serving_config:
  fixed_node_count: 0
  scaling:
    min_node_count: 1
    max_node_count: 10
    cpu_utilization_target: 80
state: STABLE

A
Create a new feature store with online serving enabled
Why wrong: Existing feature store can be updated; creation of new one is not needed.
B
Use Cloud Bigtable directly
Why wrong: Switching to Bigtable bypasses the feature store but is not a best practice.
C
Update the existing feature store to enable online serving
Online serving can be enabled by setting appropriate scaling configuration.
D
Re-import features into a new store
Why wrong: Re-importing does not fix the online serving configuration.

Full breakdown with real-world context →

Question 12easymultiple choice

Full question →

Refer to the exhibit. An auditor sees the following output from `gcloud ai models list`. What can they conclude about versioning?

Exhibit

MODEL_ID: my_model
VERSION_ID: v1
DISPLAY_NAME: my_model_v1
STATE: READY
VERSION_UPDATE_TIME: 2023-01-10T12:00:00

MODEL_ID: my_model
VERSION_ID: v2
DISPLAY_NAME: my_model_v2
STATE: READY
VERSION_UPDATE_TIME: 2023-01-15T12:00:00

A
The model is deployed on a single endpoint
Why wrong: The output does not indicate deployment status or endpoints.
B
The model has two versions with v2 being the latest
Two distinct versions are shown; v2 has a later timestamp.
C
Only the latest version is available
Why wrong: Both versions are listed with STATE: READY, so both are available.
D
The model is automatically scaled
Why wrong: The output does not provide any information about scaling.

Full breakdown with real-world context →

These PDE practice questions are part of Courseiva's free Google Cloud certification practice question bank. Courseiva provides original exam-style PDE questions with detailed explanations, topic-based practice, mock exams, readiness tracking, and study analytics.

Refer to the Exhibit Practice Questions

How to approach refer to the exhibit practice questions

Quick answer

Related PDE topic practice pages

Designing Data Processing Systems practice questions

Ingesting and Processing the Data practice questions

Storing the Data practice questions

Preparing and Using Data for Analysis practice questions

Maintaining and Automating Data Workloads practice questions

PDE fundamentals practice questions

PDE scenario practice questions

PDE troubleshooting practice questions

Practice scenarios

Refer to the exhibit. A team is trying to run a custom prediction container on Vertex AI Endpoint. They get this error when the container starts. What is the most likely cause?

Exhibit

A BigQuery query fails with the error shown in the exhibit. What is the most likely cause?

Exhibit

Refer to the exhibit. A data scientist deploys a model using this configuration. Users report that after a few hours of inactivity, the first prediction request takes over 30 seconds. What is the most likely cause?

Based on the exhibit, what is the most likely cause of duplicate rows despite using the same event_id as insertId?

Exhibit

A Dataflow pipeline as described in the exhibit has increasing lag. Which optimization is most likely to reduce the lag?

Exhibit

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

Exhibit

Refer to the exhibit. A developer sees this log entry when trying to get a prediction. What is the most likely cause?

Exhibit

Refer to the exhibit. A Cloud Build step fails when pushing a Docker image to Artifact Registry. What is the missing IAM role for the Cloud Build service account?

Exhibit

Refer to the exhibit. A data engineer sees these metrics from Cloud Monitoring for a deployed Vertex AI Endpoint. What is the most effective action to reduce latency?

Exhibit

Refer to the exhibit. The feature store 'my_fs' responds to offline queries but online serving requests fail. What is the most likely cause?

Exhibit

Refer to the exhibit. An auditor sees the following output from `gcloud ai models list`. What can they conclude about versioning?

Exhibit