--- title: "What is managed Kubernetes?" canonical: https://enum.co/learn/what-is-managed-kubernetes locale: en topic: Kubernetes publishedAt: 2026-10-15 updatedAt: 2026-10-15 author: "Max Heyer" summary: "Managed Kubernetes is Kubernetes where a provider runs the control plane, etcd and upgrades, while you deploy and run your own workloads on the worker nodes." --- # What is managed Kubernetes? Managed Kubernetes is Kubernetes where a provider runs the control plane, etcd and upgrades, while you deploy and run your own workloads on the worker nodes. After reading this you will be able to: - Explain which parts of a cluster the provider runs - List the work that stays with your team - Know what to check before choosing a managed offering ## What is managed Kubernetes? Managed Kubernetes is a Kubernetes cluster where a provider installs, runs and upgrades the control plane for you. The provider keeps the API server, etcd, the scheduler and the controller managers healthy. You get a working cluster endpoint, deploy your workloads with the usual tools, and decide how many worker nodes you need. Think of renting a flat instead of owning a house. The landlord fixes the roof, the boiler and the wiring. You still choose the furniture, lock the door and decide who gets a key. ## What does the provider run? A cluster has two halves. The control plane manages cluster state, and the worker nodes run your pods through the kubelet and a container runtime. In a managed cluster, the provider owns the control plane and most of the node lifecycle. ### Control plane and etcd The provider runs kube-apiserver, etcd, kube-scheduler and kube-controller-manager, usually as several replicas so one failed machine does not stop the API. etcd holds all cluster state, so its backups and quorum are the provider's job too. ### Upgrades and patches Kubernetes maintains release branches for the three most recent minor versions, and each minor gets about a year of patches. That means a minor upgrade at least once a year. The control plane is upgraded first, then the nodes, which are drained before their kubelet moves to the new minor version. A provider does this in a fixed order and on a schedule. Run the cluster yourself and your team plans, tests and runs every step. ### Nodes and integrations Most providers create and replace worker nodes for you, often grouped in node pools of the same size. They also connect the cluster to their load balancers, block storage and networking, so a Service of type LoadBalancer or a PersistentVolumeClaim works without extra setup. ## What stays with your team? Managed Kubernetes takes the cluster machinery off your hands, but the applications on it are still yours. Expect to own: - Manifests, Helm charts and the deploy pipeline - Resource requests, limits and how many nodes you run - RBAC, namespaces and network policies inside the cluster - Pod disruption budgets and replicas, so upgrades do not take you offline - Backups of your application data and volumes ## What should you check before choosing a provider? Ask whether each cluster gets its own control plane or shares one, and whether the provider runs upstream Kubernetes or a fork. Check what the SLA covers, which versions are supported, and whether the control plane costs extra on top of the nodes. ## How enum handles managed Kubernetes Every [enum Kubernetes Engine](/kubernetes-engine) cluster runs upstream Kubernetes with its own highly available control plane, included in the price, with a 99.9% SLA for the control plane API. enum supports the three most recent minor releases, applies patch releases and upgrades nodes one by one. Clusters are available on request today, and self-service follows later this year. ## FAQ **Do I still need Kubernetes skills with managed Kubernetes?** Yes. The provider runs the control plane, but you still write manifests, set resource limits, configure access and plan for node replacement. You need fewer people for the cluster itself, not none for the workloads. **Can I move workloads between managed Kubernetes providers?** Mostly. Manifests, Helm charts and GitOps pipelines carry over between upstream clusters. Load balancer annotations, storage classes and how pods get cloud credentials usually differ and need changes. **Is managed Kubernetes cheaper than running it yourself?** Compare total cost, not node prices. Running etcd, yearly minor upgrades and on-call for the control plane takes engineering hours every month, and those hours can cost more than the nodes.