COF-C03 Question 198
Single answer2.3 Explain monitoring and cost managementA Snowflake administrator notices that monthly compute costs have increased significantly over the last two weeks. The company uses several virtual warehouses for ELT pipelines, ad hoc analyst queries, and BI dashboards. Management wants to identify which workloads are driving the cost increase and put in place a simple mechanism to receive notifications if warehouse spending spikes again. Which action should the administrator take?
- A
Review warehouse usage in ACCOUNT_USAGE views such as WAREHOUSE_METERING_HISTORY and configure a resource monitor with notification thresholds
- B
Query LOAD_HISTORY to determine which warehouses consumed the most credits and configure Snowpipe auto-ingest notifications
- C
Review QUERY_HISTORY only, because query execution time directly represents billed credits, and set up a task to suspend warehouses after each query
- D
Use DATABASE_STORAGE_USAGE_HISTORY to identify the warehouses with the highest compute costs and create fail-safe alerts
Show answer and explanation
Correct answer: A
Explanation
For SnowPro Core, candidates should distinguish between compute monitoring, query analysis, and storage tracking. Snowflake compute costs for user-managed virtual warehouses are best analyzed with views in SNOWFLAKE.ACCOUNT_USAGE, especially WAREHOUSE_METERING_HISTORY for warehouse credit consumption over time. QUERY_HISTORY can supplement this analysis by showing which workloads ran, but it is not the primary billing view. For proactive cost management, resource monitors are the native mechanism to monitor credit consumption, trigger notifications at thresholds, and optionally suspend warehouses. In contrast, storage views such as DATABASE_STORAGE_USAGE_HISTORY help analyze storage charges, and LOAD_HISTORY focuses on data loading operations. This aligns with Snowflake best practices for monitoring usage and controlling spend.
- A. Correct.
Correct. WAREHOUSE_METERING_HISTORY in the SNOWFLAKE.ACCOUNT_USAGE schema is specifically designed to analyze credit consumption by warehouse over time, which helps identify the workloads responsible for increased compute costs. To receive alerts when spending approaches defined thresholds, a resource monitor is the appropriate Snowflake feature. Resource monitors can send notifications and optionally suspend warehouses when credit usage reaches specified limits. This directly addresses both parts of the requirement: investigate cost drivers and establish proactive monitoring.
- B. Incorrect.
Incorrect. LOAD_HISTORY tracks data loading activity, not warehouse credit consumption across general workloads such as BI, ELT, and ad hoc queries. Snowpipe notifications relate to ingestion events, not warehouse spending thresholds. This option confuses data loading observability with compute cost monitoring.
- C. Incorrect.
Incorrect. QUERY_HISTORY is useful for investigating individual queries and performance patterns, but billed credits are not derived solely from query execution time. Warehouse billing depends on warehouse runtime and size, including periods when a warehouse remains running but idle. Also, suspending a warehouse after each query is not how Snowflake cost governance is typically implemented; auto-suspend settings and resource monitors are the relevant controls.
- D. Incorrect.
Incorrect. DATABASE_STORAGE_USAGE_HISTORY is for tracking storage consumption, not compute usage by warehouse. Fail-safe is a data protection/storage concept and does not provide spend alerting for virtual warehouse credits. This option mixes storage monitoring with compute cost management.