# Scale to Zero

> Scale to zero removes an idle web deployment's replicas entirely so it costs nothing until the next request wakes it.

**Scale to zero is an autoscaling setting that lets a web deployment on Ownkube-managed compute drop to zero running replicas when it stops receiving traffic, then wake again on the next request.** Because active CPU and working-set memory only bill while a replica exists, an app at zero replicas costs nothing while it sleeps.

To turn it on, set min replicas to 0 on a web deployment's autoscaling settings, from the deployment's Settings tab. With min replicas at zero, Ownkube removes all replicas once the app has had no traffic for a while, and starts a replica again as soon as a new request arrives. That first request pays a brief [cold start](/docs/glossary/cold-start) while the replica wakes; every request after it is served normally. Scale to zero is available today for web deployments on Ownkube-managed compute, currently in beta. It isn't available for jobs, which run to completion rather than serve traffic, or for databases, which use their own scaling model and are always reserved.

Scale to zero is the sharpest edge of [active-CPU billing](/docs/glossary/active-cpu-billing) and [working-set memory](/docs/glossary/working-set-memory) metering: rather than just billing very little for an idle app, it removes the idle cost entirely. This is especially useful for low-traffic personal projects, staging environments, and preview deployments that sit unused for long stretches between visits.

- [Autoscaling](/docs/features/autoscaling)
- [On the blog: autoscaling CPU and memory](/blog/autoscale-cpu-memory)

**Don't see a feature you need?** Email [support@ownkube.io](mailto:support@ownkube.io?subject=Feature%20request). Ownkube is shaped by the teams using it and we ship what our users ask for.