Google Cloud Associate Data Practitioner - Exam-Style Scenarios¶
Original scenarios for practice. They are not vendor exam questions.
Scenario 1: A Nightly Report¶
Scenario¶
Orders and customers are loaded into BigQuery every day. The finance team needs a revenue-by-region table rebuilt at 03:30 each night. The logic is one SQL statement.
Question: Which service should schedule the rebuild?
Options: A. A BigQuery scheduled query B. A Cloud Composer environment C. A Dataproc cluster D. A Dataflow streaming job
Correct Answer: A
Explanation: - A scheduled query runs a recurring GoogleSQL statement with no extra infrastructure - The minimum interval is 5 minutes, so nightly is well within range
Why other options are wrong: - B: Composer is for multi-step dependent workflows; it is more than one statement needs - C: Dataproc runs Spark and Hadoop code, which is not required here - D: A streaming job runs continuously and is not a scheduler
Scenario 2: Moving a Large Archive¶
Scenario¶
A media company has 600 TB of footage in an on-premises file system. Their internet link would need roughly a year to upload it.
Question: What is the best way to move it to Cloud Storage?
Options: A. Storage Transfer Service over the existing link B. Transfer Appliance C. BigQuery Data Transfer Service D. Scripts using gcloud storage cp from one workstation
Correct Answer: B
Explanation: - Transfer Appliance fits when network upload would take more than a week - Data is shipped to Google and uploaded to a bucket
Why other options are wrong: - A: Still limited by the thin link - C: Loads into BigQuery, not Cloud Storage - D: Does not solve the bandwidth problem
Scenario 3: Retention Cost¶
Scenario¶
A bucket stores application logs. They are read often for 30 days, rarely until 365 days, and must be deleted after 3 years.
Question: How should the cost be controlled with the least manual work?
Options: A. Move objects by hand each quarter B. Lifecycle rules: SetStorageClass to a colder class at 30 days, Delete at about 1095 days C. Store everything in Standard and delete manually D. Copy the logs to Cloud SQL
Correct Answer: B
Explanation: - Lifecycle rules act automatically on age conditions - Colder classes cost less to store; mind minimum durations to avoid early deletion fees
Why other options are wrong: - A: Manual work and error-prone - C: Pays the highest storage price for cold data - D: Cloud SQL is not log archival storage
Scenario 4: A Failing Zone¶
Scenario¶
A Cloud SQL for MySQL database backs an order system. Management wants it to survive a zone outage with an automatic switch and the same connection address.
Question: Which configuration meets the goal?
Options: A. A read replica in the same zone B. High availability with a standby in another zone C. Daily on-demand backups only D. A scheduled query that exports data hourly
Correct Answer: B
Explanation: - An HA instance replicates synchronously to a standby in another zone and fails over automatically - Clients reconnect using the same address; expect roughly a minute of unavailability
Why other options are wrong: - A: A same-zone replica fails with the zone - C: Backups restore slowly and manually - D: Exports are not a failover mechanism
Scenario 5: Predicting Churn¶
Scenario¶
A subscription business has a BigQuery table with customer usage and a column showing who cancelled. They want a churn score for each current customer.
Question: What is the most direct approach?
Options: A. K-means clustering B. A logistic regression model trained with BigQuery ML, evaluated with ML.EVALUATE and applied with ML.PREDICT C. Export the table and sort by last login D. An ARIMA_PLUS forecast of total customers
Correct Answer: B
Explanation: - Cancelled or not is a labeled yes/no outcome, which is classification - Evaluate on held-out rows before predicting on current customers
Why other options are wrong: - A: Clustering is unlabeled grouping - C: Sorting is not a model - D: Forecasts a series, not a score per customer