GCP GCP-PDE Free Practice Questions — Page 2

Correct Answer: B. Retry the query with exponential backoff, up to a cap of 15 minutes.

Exponential backoff prevents thundering herd problems — millions of App Engine instances all retrying simultaneously would overwhelm a recovering database. Capping at 15 minutes aligns with the query frequency, so no extra staleness is introduced. Option A (restart DB) is not the frontend's responsibility. Option C (retry every second) creates massive load on a struggling DB. Option D (reduce to hourly) introduces unnecessary data staleness when the DB may recover quickly.

Correct Answer: A. Linear regression

Housing price prediction is a continuous value regression problem. Linear regression is lightweight, computationally cheap, and runs well under resource constraints. Logistic classification (B) is for categorical/binary outcomes, not price prediction. Recurrent neural networks (C) are for sequential/time-series data and are computationally expensive. Feedforward neural networks (D) can work for regression but require significantly more compute than linear regression on a constrained VM.

Correct Answer: D. Use the ROW_NUMBER window function with PARTITION by unique ID along with WHERE row equals 1.

BigQuery streaming inserts can produce at-least-once delivery, meaning duplicates are possible. Using ROW_NUMBER() OVER (PARTITION BY unique_id ORDER BY event_timestamp) and filtering WHERE row_number = 1 keeps only the first occurrence per unique ID, effectively deduplicating. Option A (ORDER BY + LIMIT 1) only works on the full result set, not per-ID. Option B (GROUP BY + SUM) distorts non-aggregatable fields. Option C (LAG function) detects consecutive duplicates but doesn't reliably deduplicate across the full dataset.

Correct Answer: D. bigquery-public-data.noaa_gsod.gsod*`

Why Option D is the correct solution: Wildcard Syntax: To query multiple tables using a wildcard, you must append an asterisk (*) to the table name base. Backticks for Project/Dataset IDs: In BigQuery, when a project ID contains hyphens (like bigquery-public-data), the entire table path must be enclosed in backticks (`). Without them, the SQL parser sees the hyphens as subtraction operators, which explains the specific error message: Expected end of statement but got "-" at line 4. Enabling _TABLE_SUFFIX: The _TABLE_SUFFIX column only becomes available for use in the WHERE clause when the FROM clause successfully identifies a wildcard pattern.

Correct Answers: B. Restrict access to tables by role.; D. Restrict BigQuery API access to approved users.; E. Segregate data across multiple tables or databases.

The principle of least privilege requires limiting access to only what's necessary. Restricting table access by role (B) uses BigQuery IAM to control who can read/write specific tables. Restricting BigQuery API access (D) prevents unauthorized users from even calling the API. Segregating data across multiple tables or databases (E) ensures users only have access to their relevant dataset. Option A (disable writes) doesn't restrict reads. Option C (encryption) is always on by default in BigQuery and doesn't control access scope. Option F (Stackdriver logging) detects violations after the fact — it doesn't enforce access control.