Why it matters
A warehouse with auto-stop turned off, or kept awake by a dashboard that refreshes on a schedule, can cost as much overnight and at weekends as during the working day.How nOps finds it
nOps reads the last 30 days of Databricks billing. A warehouse is flagged when it was billed in at least 90% of the hours and sat at its minimum level in at least 80% of off-hours. Off-hours are evenings, nights and weekends outside 13:00-24:00 UTC. The saving is what the warehouse cost in those off-hours, scaled to a month.Before you start
Know who uses the warehouse outside business hours, including scheduled queries, dashboards and alerts.How to implement
- Check what runs on the warehouse at night and at weekends, such as scheduled queries, dashboard refreshes and alerts.
- In Databricks, open SQL Warehouses, edit the warehouse, and turn on Auto stop (5 to 10 minutes of inactivity is typical).
- Move anything that keeps it awake overnight to business hours or to a separate, smaller warehouse.
- Check after a week that the warehouse now stops outside business hours.
Risks and considerations
- The first query after the warehouse stops waits while it starts again, which takes seconds on serverless and longer on Pro and Classic.
- If the business hours used by nOps do not match your team’s working hours, the saving may be lower.