Explore / Google Cloud Associate Data Practitioner Exam Prep / Practice questions
Google Cloud Associate Data Practitioner practice questions
10 questions from the bank behind this course's timed mock. Free to read and to answer, no account needed, and each has its answer and an explanation. They belong to Google Cloud Associate Data Practitioner Exam Prep.
Answer them one at a time →Read lesson 1 free
Question 1
A data engineer at a clinic is moving patient visit records from an on-premises database into BigQuery. Before upload, the pipeline drops columns that analysts do not need and fixes character encoding problems. After loading, analysts run SQL in BigQuery to join the visits with billing data and build monthly aggregates. Which data manipulation methodology does this pipeline use?
- EL with no transformation
- ETLT
- ELT
- ETL
Show the answer
Answer: B. ETLT
A light transformation happens before loading (dropping columns, fixing encodings) and the main transformation happens inside BigQuery (joins, aggregates), which is ETLT. It is not ETL, because in ETL all of the transformation is finished before loading and only clean data arrives in the target.
Question 2
A data engineer must make a one-time copy of 20 TB of log files from Microsoft Azure Blob Storage into Cloud Storage. The company can give the transfer a full 1 Gbps of bandwidth, and the copy should be managed, with retries and transfer logs. Which tool should the engineer use?
- Storage Transfer Service
- BigQuery Data Transfer Service
- Database Migration Service
- Transfer Appliance
Show the answer
Answer: A. Storage Transfer Service
At 1 Gbps (about 10 TB per day), 20 TB takes about 2 days online, and Storage Transfer Service is the managed tool for copying from Azure Blob Storage into Cloud Storage. BigQuery Data Transfer Service loads data into BigQuery, not bulk files into Cloud Storage.
Question 3
A developer exports application orders as a single JSON file that holds one large array of order objects. Each order has a nested list of line items. The developer needs to load this data into BigQuery as JSON. What must the developer change first?
- Rewrite the file as newline-delimited JSON, with one complete JSON object per line
- Compress the file with gzip so BigQuery can read it in parallel
- Flatten the nested line items into a CSV file, because BigQuery cannot load nested JSON
- Add a header row to the file that lists every key name
Show the answer
Answer: A. Rewrite the file as newline-delimited JSON, with one complete JSON object per line
BigQuery loads JSON only when it is newline-delimited, meaning one JSON object per line. The JSON can stay nested. Flattening the data to CSV is the wrong fix, because JSON supports nesting and CSV can't hold nested or repeated data at all.
Question 4
A logistics company has thousands of delivery vehicles that send GPS and sensor readings all the time. A data practitioner must enrich each reading with custom transformation logic and land the results in BigQuery within seconds. The pipeline must scale automatically as the fleet grows. Which tool should the practitioner use?
- Database Migration Service
- Dataflow
- BigQuery Data Transfer Service
- Cloud Data Fusion
Show the answer
Answer: B. Dataflow
Dataflow runs Apache Beam pipelines, which scale and handle streaming data with custom transformation logic. Cloud Data Fusion is the nearest distractor, but it is built for visual, code-free pipelines from prebuilt connectors, not real-time streaming with custom code.
Question 5
A data analyst at a London retailer creates a Colab Enterprise notebook in BigQuery Studio in the europe-west2 region. A teammate suggests attaching an existing runtime that a US team created in us-central1 so the analyst does not have to create a new one. What should the analyst do?
- Connect a runtime in europe-west2, because the runtime must be in the same region as the notebook
- Export the notebook as a .ipynb file and run it on a local Jupyter server instead
- Move the BigQuery datasets to us-central1 so they match the existing runtime
- Attach the us-central1 runtime, because a runtime can serve notebooks in any region
Show the answer
Answer: A. Connect a runtime in europe-west2, because the runtime must be in the same region as the notebook
A Colab Enterprise runtime must be in the same region as the notebook, so the analyst needs a europe-west2 runtime. Moving datasets to us-central1 does not fix the problem, because the mismatch is between the notebook and the runtime.
Question 6
A data analyst at an insurance company has structured claims data in BigQuery. The team has tried a few BigQuery ML model types and wants a more automated search for the best model. Nobody on the team wants to write custom training code. Which option fits best?
- Use AutoML on the tabular data so the managed service handles model selection and tuning
- Keep adjusting BigQuery ML model options by hand until the metrics improve
- Create a remote model over a pretrained Gemini model and call AI.GENERATE_TEXT on each claim
- Export the data and write a custom Python training pipeline
Show the answer
Answer: A. Use AutoML on the tabular data so the managed service handles model selection and tuning
AutoML suits tabular data when you need more automated model search than a few SQL options offer, and it requires no code. Tuning BigQuery ML by hand is the nearest distractor, but it does not give the managed model search the team asked for.
Question 7
A logistics company is building a new pipeline on Google Cloud. For now it processes shipment records in nightly batches. Next year it must also process the same records as a live stream. The data engineer wants one codebase for both modes, and does not want to provision or scale worker machines. Which transformation tool best meets these requirements?
- Cloud Composer running Python DAGs
- Dataproc running Spark jobs on a managed cluster
- Dataform running SQLX transformations in BigQuery
- Dataflow running an Apache Beam pipeline
Show the answer
Answer: D. Dataflow running an Apache Beam pipeline
Apache Beam uses one programming model for both batch and streaming, and Dataflow runs Beam pipelines with managed workers and autoscaling. Dataproc fits teams that already have Spark or Hadoop jobs, and nothing in this scenario points to existing Spark code.
Question 8
A data practitioner runs a streaming Dataflow job, and its job graph shows that the ParseJSON step is failing. The practitioner needs the stack trace that the pipeline's own parsing code produces when it hits a malformed record. Where in the Dataflow job UI should they look?
- Worker logs in the logs panel
- The Execution details view for the ParseJSON stage
- Job logs in the logs panel
- The Recommendations tab
Show the answer
Answer: A. Worker logs in the logs panel
Worker logs hold the output of your pipeline code, which includes stack traces from parse errors. Job logs come from the Dataflow service itself, such as start-up and scaling messages, so they will not contain the code's exception.
Question 9
A data practitioner must give a contractor read access to a Cloud Storage bucket. The access must expire automatically at the end of the quarter, so the team plans to use an IAM Condition on the bucket. The bucket currently uses fine-grained access control. What must happen before the conditional IAM binding can be applied?
- Enable uniform bucket-level access on the bucket
- Enforce public access prevention on the bucket
- Grant the contractor roles/storage.objectViewer through allAuthenticatedUsers
- Set an object ACL on each object with an expiry date
Show the answer
Answer: A. Enable uniform bucket-level access on the bucket
IAM Conditions on buckets require uniform bucket-level access, so the bucket has to be switched from fine-grained mode first. Public access prevention is a real control, but it only blocks allUsers and allAuthenticatedUsers. It does not enable IAM Conditions.
Question 10
A database administrator plans a risky schema migration on a Cloud SQL for PostgreSQL Enterprise edition instance next Friday. Automated backups use the default retention. Compliance requires that the pre-migration restore point stays available for at least six months, whatever happens to later automated backups. What should the administrator do just before the migration?
- Rely on the next automated backup and enable point-in-time recovery
- Create a read replica and leave it unchanged during the migration
- Enable high availability so the standby keeps a copy of the pre-migration data
- Take an on-demand backup of the instance
Show the answer
Answer: D. Take an on-demand backup of the instance
On-demand backups are kept until you delete them, so the restore point lasts as long as compliance requires. Automated backups with default retention (7 on Enterprise edition) rotate out within days, and PITR logs last at most 7 days on Enterprise edition.
The full timed mock
50 questions in 120 minutes, with a pass mark of 70%. Each paper is drawn from a bank of questions written for the mock alone, and these 10 are left out of it. Full access opens it, with every level of the course.
LevelUpWith is independent of the organisation that sets this exam. These questions were written for this course; they are not taken from the exam.