SnowPro Associate: Platform Question 7
Single answer● Elastic storageA retail company loads several terabytes of sales data into Snowflake every day. During a cost review, an analyst notices that the total storage billed for the database is much higher than the size of the current active tables. Investigation shows that developers frequently update and delete rows in large fact tables, and some tables are dropped and recreated during ETL testing. The team wants to understand why storage usage remains high even after data is deleted. Which explanation best describes Snowflake's elastic storage behavior in this scenario?
- A
Snowflake stores historical versions of data using micro-partitions for Time Travel and Fail-safe retention, so deleted or replaced data can continue to consume storage for a period of time.
- B
Snowflake automatically creates an additional full copy of each table in every virtual warehouse that queries it, which increases storage charges until the warehouse is suspended.
- C
Snowflake keeps deleted data only in the result cache, so storage charges increase mainly when users rerun the same queries repeatedly.
- D
Snowflake requires manual storage compaction after DELETE and UPDATE operations; until an administrator runs compaction, the storage cannot be released.
Show answer and explanation
Correct answer: A
Explanation
This scenario tests understanding of Snowflake's elastic storage model and how data lifecycle features affect billed storage. Snowflake stores table data in compressed, columnar, immutable micro-partitions in cloud storage. Because micro-partitions are immutable, UPDATE, DELETE, and similar operations write new versions rather than changing existing files in place. Older versions may still be retained for Time Travel and then Fail-safe, so storage usage can remain elevated even after data is logically removed from active tables. This is especially noticeable in ETL patterns involving frequent DML, table replacement, or drop/recreate cycles. Snowflake documentation on Time Travel, Fail-safe, and storage usage explains that historical data retention contributes to storage consumption. A key best practice is to understand retention settings and development workflows when evaluating storage costs.
- A. Correct.
Correct. Snowflake uses immutable micro-partitions in its cloud services storage layer. When data is updated or deleted, Snowflake creates new micro-partitions rather than modifying existing ones in place. Older versions can remain available during the Time Travel retention period, and then enter Fail-safe for permanent data protection. Because of this, storage billed can exceed the size of the currently active table data, especially in environments with frequent DML or object replacement.
- B. Incorrect.
Incorrect. Virtual warehouses provide compute only and do not store separate physical copies of table data. Snowflake separates compute from storage, which is a core platform design principle. Querying a table from different warehouses does not duplicate the underlying stored data.
- C. Incorrect.
Incorrect. The result cache stores query results temporarily to improve performance, but it is not the main reason for persistent storage charges related to deleted rows or dropped objects. The scenario points to retained historical data versions, not cached query output.
- D. Incorrect.
Incorrect. Snowflake does not require customers to run manual storage compaction jobs after DML operations. Storage management is automatic. The misconception here comes from traditional database systems where vacuuming or compaction may be required, but Snowflake manages micro-partitions and retention automatically.