COF-C03 Question 91
Single answer1.5 Explain Snowflake storage conceptsA data engineering team loads sales data into a Snowflake table every 15 minutes. At 3:00 PM, a faulty ETL job overwrites many rows with incorrect values. At 4:00 PM, the team notices the issue and needs to determine which Snowflake storage feature can help them query the table as it existed before the bad load without restoring from an external backup. Which Snowflake feature should they use?
- A
Time Travel, because Snowflake stores previous versions of table data for a retention period and allows querying historical data
- B
Fail-safe, because it allows users to directly query historical table versions after accidental data changes
- C
Virtual warehouse cache, because cached data blocks preserve the earlier table contents for query recovery
- D
Cloning, because Snowflake automatically creates a clone of each table version after every DML operation
Show answer and explanation
Correct answer: A
Explanation
The correct answer is Time Travel. Snowflake storage is based on immutable micro-partitions and metadata, which allows Snowflake to maintain historical versions of data for a configured retention period. This enables users to query, clone, or restore data from a previous point in time without relying on external backups. In practical terms, after accidental updates, deletes, or overwrites, Time Travel is the first feature to use. Fail-safe is different: it is a non-user-serviceable recovery window after Time Travel expires and is intended for exceptional recovery by Snowflake. Warehouse cache is a compute performance feature, not a recovery feature. Zero-copy cloning is a useful storage concept, but it must be explicitly created and is not automatically maintained for every table change. These distinctions align with Snowflake documentation on Time Travel, Fail-safe, and zero-copy cloning.
- A. Correct.
Correct. Time Travel enables access to historical data in Snowflake objects such as tables, schemas, and databases for a defined retention period. In this scenario, the team can query the table as it existed before the faulty ETL job using AT or BEFORE clauses, or potentially restore data using Time Travel. This is the primary Snowflake storage feature designed for user-accessible recovery of recent changes.
- B. Incorrect.
Incorrect. Fail-safe is a Snowflake recovery mechanism intended primarily for disaster recovery after the Time Travel retention period ends. It is not designed for direct user queries of historical data, and customers cannot use it themselves to inspect or query prior table states on demand.
- C. Incorrect.
Incorrect. Virtual warehouse cache stores query data locally for performance optimization, not for durable historical recovery. Cache contents are not a storage recovery feature and are not guaranteed to preserve prior committed table versions for business recovery scenarios.
- D. Incorrect.
Incorrect. Zero-copy cloning creates a point-in-time copy only when explicitly requested by a user. Snowflake does not automatically create clones after every DML operation. Cloning could help if a clone had been created earlier, but it is not the built-in feature intended for accessing prior versions after accidental updates.