Skip to content

Kubernetes

The control plane is ours. The cluster is yours.

A managed Kubernetes control plane — etcd, the API server and the upgrades — run for you, with worker nodes billed as ordinary servers. Available in Amsterdam, Atlanta, Mumbai, Jakarta, Dallas, Frankfurt, Sao Paulo, Washington, Osaka, Los Angeles, London, Chennai, Madrid, Melbourne, Miami, Milan, Tokyo, New York, Chicago, Paris, Seattle, Silicon Valley, Singapore, Stockholm, Sydney, Toronto. The standard control plane costs nothing; you pay for the nodes.

Price

What it costs, and what that includes.

The control plane is what is priced here. Worker nodes are ordinary servers from the VPS catalogue and are billed at those prices — a cluster's real cost is almost entirely the nodes.

  • K8S-STANDARDNo charge for the component

    Standard control plane — no charge for the control plane itself; you pay for the worker nodes

    Included

  • K8S-HA

    High availability control plane, with the worker nodes billed separately

    $151.90 / mo

  • K8S-ENTERPRISE

    Enterprise control plane, with the worker nodes billed separately

    $540.90 / mo

Runs in: Amsterdam, Atlanta, Mumbai, Jakarta, Dallas, Frankfurt, Sao Paulo, Washington, Osaka, Los Angeles, London, Chennai, Madrid, Melbourne, Miami, Milan, Tokyo, New York, Chicago, Paris, Seattle, Silicon Valley, Singapore, Stockholm, Sydney, Toronto.

How you get it: set up by hand today, within business hours — someone reads the order before anything is built, and you get the details by email. We are writing the automation; until it is done and tested, this is the honest timing.

What you get

The parts that are ours.

Upgrades without a maintenance window of yours

Control-plane patching and Kubernetes version upgrades are handled for you. Node pools are rolled when you choose, so the part that touches your workloads stays on your calendar.

A kubeconfig and nothing else to install

You get a kubeconfig and a cluster that answers kubectl. Ingress controllers, cert-manager, monitoring and whatever else your stack expects are yours to install — we do not pre-install opinions.

Nodes priced like servers

Worker nodes come from the same catalogue as everything else here, at the same prices, in the same cities. No cluster surcharge on the machines.

Honesty

When you should not buy this.

Kubernetes is not a better way to run one application on one server — it is a way to run many, by a team that already knows it. If nobody on your side can debug a CrashLoopBackOff at 3am, a VPS with docker compose will serve you better and cost less, and we will tell you so rather than sell you a cluster. Come back to this when you have the second team, the second service, or a real reason to schedule work across machines.

Questions

The ones worth asking first.

What exactly costs nothing?
The standard control plane. You pay for worker nodes, and for a high-availability control plane if you choose one — the price is in the table above. Nothing is free that costs us money; the supplier does not charge for the standard control plane, so neither do we.
Should I take the high-availability control plane?
If a control-plane restart would be an incident for you, yes. In practice it matters when your cluster is doing continuous work — autoscaling, frequent deploys, operators reconciling — because those stop while the API server is away, even though running pods keep serving. For a small cluster that deploys a few times a week, the standard plane is honest value.
Do you manage what runs inside the cluster?
No. The control plane and the nodes are ours to keep alive; your manifests, images, ingress and application health are yours. If you want someone to run the workloads too, that is a managed-hosting conversation rather than this product.
How is a cluster set up today?
By hand, within business hours, after a short conversation about node sizes and how many. Clusters are the one product where that is arguably a feature — the sizing questions are worth asking out loud before anything is built.