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

This page has 10 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 →

In the Vertex AI Pipeline component YAML exhibit, the component is designed to evaluate a model and produce metrics. If the threshold_accuracy is set to 0.85, what is the expected behavior of this component?

Exhibit

Refer to the exhibit.

```
# Vertex AI Pipeline component YAML
name: model-evaluation
inputs:
  model_path:
    type: String
  test_data_path:
    type: String
  threshold_accuracy:
    type: Float
    default: 0.85
outputs:
  evaluation_metrics:
    type: Metrics
implementation:
  container:
    image: gcr.io/my-project/eval:latest
    args: [
      --model_path, {inputValue: model_path},
      --test_data_path, {inputValue: test_data_path},
      --threshold_accuracy, {inputValue: threshold_accuracy},
      --output_path, {outputPath: evaluation_metrics}
    ]
```

A
It will output the evaluation metrics, and the pipeline can use them for conditional decisions
The component emits evaluation metrics as pipeline artefacts rather than enforcing the threshold itself; downstream conditional logic, such as a threshold comparison step, consumes those metrics to decide whether to promote the model. Setting threshold_accuracy to 0.85 therefore supplies the comparison value, not an automatic pass/fail gate.
B
It will deploy the model if the accuracy meets the threshold
Why it fails: The component only evaluates and emits metrics; model deployment is a separate pipeline step. It is tempting because thresholds commonly gate promotion, but this YAML's outputs are metrics, not deployment actions, so meeting the threshold does not itself deploy the model.
C
It will ignore the threshold_accuracy input if not provided
Why it fails: The component compares the produced metric against threshold_accuracy and emits a boolean outcome; an unset input would use its default, not be ignored. It is tempting because optional inputs can be omitted, but the YAML defines a default value, so the threshold still governs evaluation rather than being skipped.
D
It will fail if the model accuracy is below 0.85
Why it fails: The component outputs a boolean metric indicating whether accuracy meets the threshold; it does not raise a pipeline failure. It is tempting because threshold breaches often trigger alerts, but this component's contract is to emit the comparison result, leaving failure handling to downstream conditional logic.

Full breakdown with real-world context →

Question 2hardmultiple choice

Full question →

Refer to the exhibit. A BigQuery dataset is shared with the group 'analysts@example.com' using the IAM policy shown. A user who is a member of this group reports that they cannot run queries on the dataset, though they can see the tables. What is the most likely reason?

Exhibit

Refer to the exhibit.

```json
{
  "bindings": [
    {
      "role": "roles/bigquery.dataViewer",
      "members": [
        "group:analysts@example.com"
      ]
    }
  ]
}
```

A
The group needs the 'roles/bigquery.jobUser' role at the project level.
BigQuery separates metadata visibility from query execution: dataset-level roles such as `roles/bigquery.dataViewer` let the user list tables, but running jobs requires `roles/bigquery.jobUser` granted at the project level. Because the group only holds dataset-scoped access, the user can see tables yet cannot execute queries, satisfying the stem's constraint.
B
The user is using an incorrect client library version.
Why it fails: A client library version affects how requests are constructed, not whether the identity holds bigquery.dataViewer versus bigquery.user; the user can already list tables, so connectivity works. Library upgrades matter when an API method or authentication flow is unsupported by an outdated release.
C
The user's account is not activated in the group membership.
Why it fails: Group membership is evaluated from the directory, not activated per user; an unactivated membership would also prevent the table visibility the user already has. Deactivated accounts are the right diagnosis when a user cannot authenticate or appears absent from group-based access entirely.
D
The dataset has an organization policy that denies query access.
Why it fails: An organisation policy denying query access would also block the table listing the user currently sees, so it cannot explain visibility without query rights. Such policies are correct when you must restrict access by resource hierarchy regardless of IAM, for example enforcing domain-restricted sharing.

Full breakdown with real-world context →

Question 3mediummultiple choice

Full question →

Refer to the exhibit. A Dataflow streaming pipeline subscribes to this Pub/Sub subscription. The pipeline occasionally takes more than 10 seconds to process a message. Which behavior will occur?

Exhibit

Refer to the exhibit.

Cloud Pub/Sub subscription configuration:

{
  "name": "projects/my-project/subscriptions/my-sub",
  "topic": "projects/my-project/topics/my-topic",
  "pushConfig": {},
  "ackDeadlineSeconds": 10,
  "messageRetentionDuration": "86400s",
  "expirationPolicy": {
    "ttl": "604800s"
  },
  "enableMessageOrdering": false,
  "retryPolicy": {
    "minimumBackoff": "10s",
    "maximumBackoff": "600s"
  },
  "deadLetterPolicy": {
    "deadLetterTopic": "projects/my-project/topics/dead-letter-topic",
    "maxDeliveryAttempts": 5
  }
}

A
The message will be sent to the dead letter topic immediately.
Why it fails: Pub/Sub redelivers a message when its acknowledgement deadline expires; exceeding ten seconds triggers redelivery, not immediate dead-lettering. Dead-lettering occurs only after the configured maximum delivery attempts are exhausted, so a slow message is retried rather than discarded at once.
B
The message will be retried with exponential backoff as per retry policy.
Why it fails: Pub/Sub retry policies apply only to acknowledged failures; a message still within its ack deadline is neither redelivered nor backed off, so the pipeline's slow processing simply continues until the deadline lapses. Retry policy with exponential backoff suits push subscriptions where the endpoint returns an error.
C
The message will be redelivered after 10 seconds if not acknowledged.
The subscription's acknowledgement deadline is 10 seconds, so a message still unacknowledged when that window expires becomes eligible for redelivery. Because processing occasionally exceeds 10 seconds, the pipeline may receive duplicates, requiring idempotent handling or an extended deadline to avoid repeated processing.
D
The message will be dropped after 10 seconds due to expiration policy.
Why it fails: Expiration policy governs how long Pub/Sub retains an unacknowledged message before discarding it, typically days, not the 10-second ack deadline. Dropping after 10 seconds describes the ack deadline expiring, which triggers redelivery rather than deletion. Expiration policy is the right control for discarding stale backlog.

Full breakdown with real-world context →

Question 4mediummultiple choice

Full question →

Refer to the exhibit. What is the cause of this error?

Network Topology

A
The machine type flag is only used during model deployment, not endpoint creation
The machine type is a deployment-time property of the Model resource, not the Endpoint. Passing it during endpoint creation is rejected because the API expects no such field there, which is exactly the constraint the error reflects.
B
The endpoint name already exists
Why it fails: The failure stems from the request referencing an endpoint that does not exist or is misnamed, not from a duplicate name conflict. It is tempting because endpoint names must be unique within a project, so a duplicate-name error would occur when creating an endpoint that already exists.
C
The user must specify a model name
Why it fails: The error is produced by a malformed or unsupported request payload rather than a missing model name, which the platform supplies by default. It is tempting because some prediction APIs do require an explicit model identifier, and omitting it would then cause a validation failure.
D
The region is missing
Why it fails: The error arises from the specified region not being enabled or available for the service, so requests to that regional endpoint fail. It is tempting because region misconfiguration does genuinely cause endpoint and connectivity errors, and selecting a supported region would be the fix in that scenario.

Full breakdown with real-world context →

Question 5mediummultiple choice

Full question →

Refer to the exhibit. This log entry was generated by Vertex AI Model Monitoring for a production model. What should the data engineer do to address this issue?

Exhibit

{
  "resource": {"type": "ai_platform_endpoint", "labels": {"endpoint_id": "123"}},
  "severity": "ERROR",
  "jsonPayload": {
    "feature_name": "age",
    "monitoring_type": "prediction_drift",
    "drift_score": 0.85,
    "threshold": 0.7
  }
}

A
Increase the drift threshold to 0.9 to suppress alerts
Why it fails: Raising the threshold to 0.9 hides genuine distribution shift rather than resolving it, leaving the model serving predictions on drifted data. Thresholds exist to trigger investigation when drift exceeds tolerance; suppressing alerts is defensible only when the observed drift is confirmed benign and expected.
B
Retrain the model with more recent data
The monitoring log indicates training-serving skew or drift, where live feature distributions diverge from those the model learned. Retraining on recent production data realigns the model with current patterns, restoring prediction accuracy without altering the serving pipeline.
C
Deploy a new model version trained on the original dataset
Why it fails: Retraining on the original dataset reproduces the same feature distribution the model already learned, so incoming drift persists. Retraining is the right response to concept drift or degraded accuracy, but here it must use recent production data reflecting the shifted distribution, not the stale baseline.
D
Disable monitoring for the 'age' feature
Why it fails: Disabling monitoring for 'age' removes the signal detecting distribution shift, so future drift in that feature goes unnoticed. Feature-level monitoring can legitimately be disabled for features confirmed irrelevant to predictions, but 'age' is typically a meaningful model input, making suppression inappropriate.

Full breakdown with real-world context →

Question 6easymultiple 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
roles/artifactregistry.writer grants the push permission Cloud Build requires to upload image layers to the repository. Without it, the step fails with a permission denied error, so this role directly resolves the failing push described in the exhibit.
B
roles/containerregistry.admin
Why it fails: roles/containerregistry.admin applies to the legacy Container Registry service, not Artifact Registry repositories, so the push is still denied. It is tempting because it is the obvious image-push role, and would be correct if the build targeted gcr.io rather than an Artifact Registry repository.
C
roles/storage.objectCreator
Why it fails: roles/storage.objectCreator grants writes to Cloud Storage buckets, not Artifact Registry repositories, so the Docker push remains unauthorised. It is tempting because Artifact Registry historically stored images in GCS-backed buckets, and would be correct if the build were uploading objects to a Cloud Storage bucket instead.
D
roles/cloudbuild.builds.editor
Why it fails: roles/cloudbuild.builds.editor governs Cloud Build job creation and execution, not Artifact Registry repository writes, so the push still fails. It is tempting because it is the natural Cloud Build role, and would be correct if the failure were in triggering or managing builds rather than authenticating the image push.

Full breakdown with real-world context →

Question 7mediummultiple choice

Full question →

Refer to the exhibit. A team uses this Cloud Build configuration to deploy a service to Cloud Run. The deployment step fails with a 'Permission denied' error. What is the most likely cause?

Network Topology

A
The Dockerfile is missing from the repository.
Why it fails: A missing Dockerfile produces a build-step failure such as 'no such file', not a permission error during deployment. Cloud Build would fail before reaching the Cloud Run deploy step. This is tempting because build context problems are common, and it would be correct if the build step itself reported the file could not be found.
B
The Docker image tag is missing or malformed.
Why it fails: A malformed or absent image tag causes an invalid image reference error, not 'Permission denied'. Tag issues surface as manifest-not-found or invalid-format messages. This is tempting because tag mistakes are frequent in Cloud Build configs, and it would be the answer if the error named the image reference rather than authorisation.
C
The region 'us-central1' is incorrect for Cloud Run.
Why it fails: An incorrect region yields a region-not-found or unsupported-location error, not 'Permission denied'. Cloud Run accepts us-central1. This is tempting because region mismatches do break deployments, and it would be correct if the error cited an unavailable location or the service was deployed to a region the caller cannot access.
D
The Cloud Build service account does not have the Cloud Run Admin role.
Cloud Build executes deployments using its dedicated service account, which requires the Cloud Run Admin role to create or update services. Without that IAM binding, the deployment step is rejected with 'Permission denied' even though the configuration syntax itself is valid.

Full breakdown with real-world context →

Question 8mediummultiple choice

Full question →

Refer to the exhibit. A Dataflow pipeline is failing intermittently with the shown error. Which step should the team take to ensure data quality and prevent such errors?

Exhibit

Refer to the exhibit.
{
  "insertId": "abc123",
  "jsonPayload": {
    "message": "Error processing element: expected integer at field 'temperature', got string 'hot'",
    "workerId": "worker-5",
    "step": "ParseAndValidate"
  },
  "resource": {
    "type": "dataflow_step",
    "labels": {
      "job_id": "job-1234",
      "step_id": "s2"
    }
  }
}

A
Increase the number of workers to process the data faster.
Why it fails: Adding workers raises parallel throughput but does not inspect record contents, so the same malformed input still triggers the failure; the error is data-driven, not capacity-driven. Scaling is tempting because intermittent failures often look like resource contention, and would be correct for genuine hot-keying or backlog-driven latency.
B
Add a monitoring alert on the 'system_lag' metric.
Why it fails: A system_lag alert only notifies after backlog accumulates; it detects the symptom rather than validating records, so malformed data still reaches the pipeline and the error recurs. Alerting is tempting because it is the usual operational response to intermittent failures, and would be right for latency or throughput SLO breaches.
C
Use a strongly typed schema for the PCollection and let Beam automatically reject malformed data.
Why it fails: A schema rejects records that violate declared types at encoding time, but Beam does not silently drop them; the pipeline still fails on the offending element unless a dead-letter path handles it. Schemas are tempting because they enforce structure, and would be right when downstream consumers need typed fields.
D
Modify the pipeline to handle parsing failures by sending invalid records to a dead letter queue.
Routing unparseable records to a dead letter queue isolates malformed data so the pipeline continues processing valid rows, preventing the intermittent failures caused by parsing errors while preserving invalid records for later inspection and remediation.

Full breakdown with real-world context →

Question 9mediummultiple choice

Full question →

The exhibit shows a Cloud Logging query result. A data engineer sees this log for a streaming Dataflow job. What is the most likely cause?

Exhibit

Refer to the exhibit.

```
resource.type="dataflow_step"
resource.labels.job_id="2023-01-01_000000-12345678"
"worker pool exhausted"
```

A
The job is experiencing network latency.
Why it fails: Network latency would surface as elevated RPC retry counts or watermark lag, not the logged error. The tempting pull is that streaming Dataflow pipelines do depend on steady throughput to Google Cloud services, so latency is a plausible suspect when a job stalls. Here the log names a specific failure the correct option addresses directly.
B
The job is using too much memory per worker.
Why it fails: Excessive per-worker memory triggers out-of-memory worker restarts, not the watermark or backlog symptoms shown. It is tempting because memory tuning genuinely resolves throughput problems when workers spill or crash under load, making it the right fix for a job whose logs show OOM errors rather than the exhibited pattern.
C
The job has insufficient permissions to scale.
Why it fails: Insufficient permissions would surface as an authorisation error, not the exhibited log pattern. Scaling in Dataflow is governed by autoscaling and worker resource metrics, so IAM roles cannot produce this symptom. Permission failures are the right diagnosis when a job cannot read its source or write to its sink at all.
D
The job has reached the maximum number of workers allowed by the project quota.
Dataflow caps worker count at the project quota; once the autoscaler requests more workers than permitted, the job stalls and logs quota-related errors rather than processing backlog. This matches the observed log for a streaming pipeline that cannot scale further.

Full breakdown with real-world context →

Question 10hardmultiple 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 it fails: More workers may help but the streaming insert bottleneck remains.
C
Use global windows instead of fixed windows
Why it fails: Global windows would increase latency as data waits until end.
D
Add additional ParDo transforms
Why it fails: More transforms add processing overhead, increasing lag.

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

Storing the Data practice questions

Designing Data Processing Systems practice questions

Ingesting and Processing 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

In the Vertex AI Pipeline component YAML exhibit, the component is designed to evaluate a model and produce metrics. If the threshold_accuracy is set to 0.85, what is the expected behavior of this component?

Exhibit

Refer to the exhibit. A BigQuery dataset is shared with the group 'analysts@example.com' using the IAM policy shown. A user who is a member of this group reports that they cannot run queries on the dataset, though they can see the tables. What is the most likely reason?

Exhibit

Refer to the exhibit. A Dataflow streaming pipeline subscribes to this Pub/Sub subscription. The pipeline occasionally takes more than 10 seconds to process a message. Which behavior will occur?

Exhibit

Refer to the exhibit. What is the cause of this error?

Refer to the exhibit. This log entry was generated by Vertex AI Model Monitoring for a production model. What should the data engineer do to address this issue?

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 team uses this Cloud Build configuration to deploy a service to Cloud Run. The deployment step fails with a 'Permission denied' error. What is the most likely cause?

Refer to the exhibit. A Dataflow pipeline is failing intermittently with the shown error. Which step should the team take to ensure data quality and prevent such errors?

Exhibit

The exhibit shows a Cloud Logging query result. A data engineer sees this log for a streaming Dataflow job. What is the most likely cause?

Exhibit

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

Exhibit