ADA-C01 Question 427
Single answerMonitor Notebooks consumptionA Snowflake administrator is asked to identify which teams are driving the highest Snowflake Notebook compute usage this month so the organization can charge back costs. The company tags warehouses by team, but Notebook users often switch between warehouses during development. The administrator needs a reliable way to monitor Notebook consumption and attribute the spend as accurately as possible. Which approach should the administrator use?
- A
Query account usage views for warehouse metering and correlate the warehouses used by Notebook sessions over time, then aggregate consumption based on the associated warehouse usage
- B
Use QUERY_HISTORY only, because every Notebook action maps directly to a single SQL statement with complete cost data attached to the query
- C
Review only the NOTEBOOK object metadata in the account, because Notebook objects store total compute consumption by owner and by execution session
- D
Query STORAGE_USAGE to determine which Notebooks consumed the most compute, because Notebook runtime consumption is recorded as storage for the Notebook files
Show answer and explanation
Correct answer: A
Explanation
To monitor Snowflake Notebook consumption, administrators should focus on the compute resource that powers Notebook execution. In Snowflake, Notebook work runs using a selected warehouse, so the most reliable way to measure and attribute cost is to analyze warehouse usage and correlate it with Notebook activity over time. This is especially important when users switch warehouses within the month, because chargeback must follow the actual warehouse consumption rather than simply the Notebook owner or object metadata. QUERY_HISTORY can help with operational visibility, but warehouse metering and related account usage views are the authoritative basis for compute cost analysis. As a best practice, combine warehouse metering data with governance controls such as warehouse tagging, warehouse assignment standards, and periodic review of Notebook usage patterns. This aligns with Snowflake guidance that compute costs are fundamentally associated with the compute resources executing workloads, while storage views measure storage rather than runtime consumption.
- A. Correct.
Correct. Snowflake Notebooks consume compute through the warehouse selected for the Notebook session. To monitor Notebook consumption for chargeback, an administrator should use usage/account usage telemetry related to warehouse consumption and correlate that with Notebook activity and warehouse assignment over time. This is the most practical approach because costs are incurred by the compute resource actually executing the Notebook workload. If users switch warehouses, attribution must follow the warehouse actually used during those periods rather than relying only on object ownership.
- B. Incorrect.
Incorrect. QUERY_HISTORY is useful for understanding SQL activity generated from Notebook work, but it is not sufficient by itself for complete Notebook consumption attribution. Not all Notebook activity should be reduced to a single statement-level cost model, and query history does not directly provide a definitive, complete per-Notebook cost record. Warehouse consumption is billed at the warehouse level, so metering data is required for accurate chargeback.
- C. Incorrect.
Incorrect. Snowflake Notebook objects do not provide a built-in object metadata field that stores total compute consumption by owner or by session in the way described here. This distractor reflects a common misconception that all Snowflake object types expose direct chargeback metrics. In practice, administrators need to derive Notebook cost from compute usage telemetry rather than relying on Notebook object metadata alone.
- D. Incorrect.
Incorrect. STORAGE_USAGE is related to storage consumption, not warehouse compute consumption. Notebook runtime and interactive execution costs are tied to compute resources, primarily warehouses, not to storage accounting for Notebook files. Someone might choose this if they confuse persisted Notebook artifacts with the cost of running code interactively.