Skip to main content

Building Admin Monitoring Dashboards

As a Snowflake Administrator, you are responsible for ensuring that the platform is secure, performant, and cost-efficient. You cannot achieve this by manually running SQL queries every day.

You must build automated Monitoring Dashboards that visualize your Account Usage data and surface anomalies immediately.

Topics Covered​

  • Key Administrative Metrics
  • Building Snowsight Dashboards
  • Example Dashboard Queries (Costs, SLAs, Security)
  • External BI Integration

1. Key Administrative Metrics​

A professional operations dashboard should track three pillars:

  1. Financial Operations (FinOps): Which warehouses are burning the most credits? Are we over our monthly budget?
  2. Performance Operations (DataOps): Are critical ETL tasks failing? Are business queries taking longer than their defined SLA (Service Level Agreement)?
  3. Security Operations (SecOps): Are there brute-force login attempts? Are unauthorized users attempting to access restricted tables?

2. Building Snowsight Dashboards​

Snowflake provides a native visualization tool called Snowsight. It allows you to write SQL queries and instantly turn them into bar charts, line graphs, and scorecards.

(To build these dashboards, you should use a role with access to the SNOWFLAKE database, such as ACCOUNTADMIN or an explicitly granted reporting role. Verify your identity by running SELECT CURRENT_USER();)

To create a dashboard:

  1. Log into the Snowflake Web UI.
  2. Navigate to Dashboards -> + Dashboard.
  3. Create new "Tiles" (SQL queries) and arrange them.

3. Hands-On: Example Dashboard Queries​

Here are three essential queries you should include in your Admin Dashboard.

Tile 1: Warehouse Credit Burn Rate (FinOps)​

This query calculates how many credits each warehouse has consumed over the last 30 days. You would visualize this as a Bar Chart.

USE ROLE ACCOUNTADMIN;

SELECT
WAREHOUSE_NAME,
SUM(CREDITS_USED) AS TOTAL_CREDITS,
-- Assuming a rough cost of $3.00 USD per credit (convert to INR for local reporting if needed)
SUM(CREDITS_USED) * 3.00 AS ESTIMATED_COST_USD
FROM SNOWFLAKE.ACCOUNT_USAGE.WAREHOUSE_METERING_HISTORY
WHERE START_TIME >= DATEADD('day', -30, CURRENT_TIMESTAMP())
GROUP BY WAREHOUSE_NAME
ORDER BY TOTAL_CREDITS DESC;

Tile 2: Task Failure Rate (DataOps)​

If the overnight ETL tasks for your Indian logistics company fail, the morning reports will be blank. This query finds all failed tasks in the last 24 hours. Visualize this as a Data Table.

SELECT 
NAME AS TASK_NAME,
ERROR_MESSAGE,
SCHEDULED_TIME,
COMPLETED_TIME
FROM TABLE(INFORMATION_SCHEMA.TASK_HISTORY(
SCHEDULED_TIME_RANGE_START => DATEADD('day', -1, CURRENT_TIMESTAMP())
))
WHERE STATE = 'FAILED'
ORDER BY SCHEDULED_TIME DESC;

Tile 3: Top 10 Longest Queries (DataOps)​

This query identifies the exact queries that are taking the longest to execute. These are prime candidates for optimization (e.g., adding clustering keys or rewriting the SQL).

SELECT 
QUERY_ID,
USER_NAME,
WAREHOUSE_NAME,
EXECUTION_TIME / 1000 AS EXECUTION_SECONDS,
BYTES_SCANNED / 1024 / 1024 / 1024 AS GIGABYTES_SCANNED,
SUBSTR(QUERY_TEXT, 1, 50) || '...' AS QUERY_SNIPPET
FROM SNOWFLAKE.ACCOUNT_USAGE.QUERY_HISTORY
WHERE START_TIME >= DATEADD('day', -1, CURRENT_TIMESTAMP())
AND EXECUTION_TIME > 0
ORDER BY EXECUTION_TIME DESC
LIMIT 10;

4. Sharing and Scheduling Dashboards​

Once your dashboard is complete, you do not want to force the VP of Engineering to log into Snowflake to check it.

  1. Read-Only Sharing: You can share a Snowsight dashboard with other Snowflake roles. The viewers can see the charts and the data, but they cannot modify the underlying SQL queries.
  2. Scheduled Emails (Coming Soon): Snowflake is actively developing native features to schedule automated PDF/email snapshots of dashboards.

5. External BI Integration​

While Snowsight is excellent for quick admin tasks, massive enterprises usually use dedicated BI tools (Tableau, PowerBI, Looker) for reporting.

You can connect Tableau directly to the SNOWFLAKE.ACCOUNT_USAGE schema. This allows you to:

  • Blend Snowflake cost data with Azure/AWS infrastructure costs for a unified FinOps dashboard.
  • Create complex, interactive drill-down reports for departmental chargebacks.
  • Set up custom email alerts (e.g., "Alert me if the Data Science warehouse burns more than 50 credits in a single day").

Summary​

  • Automated Monitoring Dashboards are essential for tracking FinOps, DataOps, and SecOps.
  • Snowsight allows you to rapidly build charts directly inside the Snowflake UI.
  • Essential metrics include: Warehouse Credit Consumption, Task Failures, and Long-Running Queries.
  • For enterprise-wide visibility, connect the ACCOUNT_USAGE schema to external tools like Tableau or PowerBI.