Resource Parking
Scheduled shutdown — AutoStopping, off-hours parking, hibernation, and scale-to-zero in one control plane, with unsafe-schedule detection.
At a glance
| Route | /resource-parking |
| Group | Govern |
| Page permission | resource_parking |
What it is
A unified control plane for scheduled shutdown: AutoStopping, off-hours parking, hibernation, and scale-to-zero. Resources that are only needed during working hours are stopped outside them, on a schedule, automatically.
Who it is for
Platform engineers and FinOps practitioners targeting the largest and most reliable saving available in most estates — non-production infrastructure running twenty-four hours a day for a forty-hour working week.
How it works
A schedule defines which resources to park, when to stop them, and when to restart them. Candidate resources are matched against live inventory, and the module explicitly reports when no live inventory matches — rather than silently creating a schedule that will never act.
Unsafe-schedule detection
Schedules are validated before activation, and unsafe ones are detected and flagged — a schedule that would stop production infrastructure, for example.
This addresses the one real risk in this module: an over-broad selector that accidentally matches production.
Features
- AutoStopping, off-hours parking, hibernation, and scale-to-zero in one control plane
- Schedule creation, editing, deletion, and manual triggering
- Live inventory matching, with explicit reporting when no resources match
- Schedule validation with unsafe-schedule detection
- Preview mode and blocked-schedule states
- Approval routing for live actions
- Per-schedule run history
- Parking recommendations identifying candidate resources
- Support matrix showing which resource types can be parked per provider
- Summary statistics on savings delivered
How to use it
Review Recommendations first
The module identifies parking candidates from utilisation patterns rather than requiring you to nominate them.
Check the support matrix
Confirm the target resource type can be parked on the relevant provider before designing a schedule around it.
Create a schedule in preview mode
And confirm it matches the intended resources and only those.
Resolve any unsafe-schedule warnings before proceeding
They exist to catch production infrastructure caught by an over-broad selector. Do not override them casually.
Check run history after the first cycle
Confirm both the stop and the restart executed. The restart is the half people forget to verify.
Widen coverage gradually
Across non-production environments, reviewing savings statistics monthly.
Why it matters
Non-production infrastructure typically runs continuously while being used for roughly a quarter of the week. That makes scheduled shutdown the highest-certainty saving available in most estates — commonly sixty to seventy percent of non-production compute cost.
It is also unusually safe, because the action is reversible by design and nothing is destroyed.
Connects to
- Shares execution machinery with Optimization and Autopilot
- Actions route through Approvals
- Candidates identified from Inventory and Opportunity Queue
- Protected tags are honoured throughout