Scheduling Azure SQL Database
Can ZopNight schedule Azure SQL Database?
Azure SQL Database pauses only on the Serverless compute tier. Provisioned DTU and vCore databases have no pause at all. A calendar pause beats the native auto-pause whenever 1 lingering connection keeps the idle detector from ever firing, and a paused database bills storage while compute drops to zero.
How the stop works
| Field | Value |
|---|---|
| Behaviour | paused and resumed, but only for databases on the Serverless compute tier; provisioned-tier databases cannot be paused. |
Example schedules
- 0 8 * * 1-5 — Business Hours Start: Start at 8:00 AM on weekdays
- 0 18 * * 1-5 — Business Hours Stop: Stop at 6:00 PM on weekdays
- 0 22 * * * — Night Shutdown: Stop at 10:00 PM every day
- 0 6 * * 1-5 — Morning Startup: Start at 6:00 AM on weekdays
- 0 20 * * 5 — Weekend Shutdown: Stop at 8:00 PM on Friday
- 0 7 * * 1 — Weekend Startup: Start at 7:00 AM on Monday
The tier decides whether pause exists
Pause is a Serverless-tier capability, full stop. Provisioned databases, whether DTU or vCore, run continuously by design and no schedule can pause them; their cost lever is scaling down, a different operation with different latency. The first eligibility check on any Azure SQL schedule is therefore the compute tier, and a fleet audit usually finds a mix.
Databases inside elastic pools are excluded for a different reason: the pool bills as a unit, and pools have no pause operation at all.
Auto-pause versus the calendar
Serverless already auto-pauses after an idle delay of at least 1 hour. The catch is the definition of idle. A connection pool minimum of 1, a monitoring probe or an abandoned SSMS tab is enough to reset the detector forever, and the database bills compute all night while technically “auto-pausing”. The scheduled pause is deterministic where the detector is hopeful. Databases that reliably auto-pause need no schedule; databases that never quite manage it are the target audience.
Resume happens on first touch
A paused database wakes when a connection arrives, with a warm-up delay around a minute, and the first connection attempts fail while it climbs. Applications need retry logic on connect, which they should have anyway. A scheduled morning resume ahead of business hours converts that user-facing stumble into an invisible one.
Paused still bills storage
Compute billing reaches zero during a pause; storage billing does not, covering data and backups at their normal rates. The saving on a lightly-used serverless database is the compute line, which for a database idle 16 hours a day is the majority, though never the whole bill.
Configurations that block pausing
Certain features pin a serverless database awake; long-term geo-replication arrangements are the classic case. If the pause never seems to stick, check the feature list before blaming the schedule; the database may be contractually unpausable in its current shape.