Managed Kubernetes
Upstream Kubernetes clusters you can create from one page, scale from the console and manage as code.
Upstream Kubernetes, built inside your own private network
Choose a location, a node size, a network plugin and a version. VPSie creates the virtual machines, bootstraps the cluster with kubeadm, joins the workers, installs the storage and cloud-controller integrations, and gives you a kubeconfig.
- Standard Kubernetes, so existing manifests, Helm charts and GitOps tools work unchanged
- Calico or Cilium container networking, chosen at creation
- Nodes sit on private addresses in a VPC you own
- No separate fee for the control plane or cluster management
How a cluster is built
Every cluster is made of VPSie servers, so it uses the same account, projects and networks.
Control plane
One or more masters run the API server, scheduler, controller manager and etcd. With several masters, an HAProxy load balancer publishes a single API endpoint on port 6443, so your kubeconfig does not change when masters are added.
- Scale down is refused if it would break etcd quorum
Worker node groups
Each group has its own plan size, so general-purpose and memory-heavy workloads share one cluster. Add or remove workers at any time.
- Add a new group with a different size
Built-in integrations
The VPSie CSI driver creates block volumes for PersistentVolumeClaims, and the VPSie cloud controller manager links the cluster to the platform. Their extended features are in Preview, listed below.
- Basic integrations are set up when the cluster is created
Create a cluster on one page
Region, plan and counts
Pick a data center that offers Kubernetes, then master and worker counts and the first node group name.
VPC, CNI and version
Select or create a VPC, choose Calico or Cilium and a Kubernetes release from the console catalog.
Name, DNS and project
Accept the generated name or set your own, optionally bind a private DNS domain, and pick a project.
Then create the cluster. Provisioning, bootstrap and add-on installation run on their own. Download the kubeconfig when it is ready.
Day-2 operations without SSH to the nodes
| Operation | What you get |
|---|---|
| Scale the control plane | Add or remove masters, with quorum protection |
| Scale workers | Add or remove workers in any group |
| Upgrade | Apply patch updates or move to a newer minor version from the console |
| Certificates | Expiry date shown on the cluster; checked and renewed automatically |
| Access | Download the kubeconfig and change the API port |
| Protection | Lock a cluster against deletion; deleting needs a reason |
| Organize | Move clusters between projects and export lists to CSV or PDF |
Autoscaling and add-ons
Scaling by hand is available today. Automatic node scaling is in preview.
Node autoscaler
Adds nodes when pods cannot be scheduled and drains and removes under-used nodes while respecting PodDisruptionBudgets. NodeGroup resources, Prometheus metrics and a Helm chart are included.
CSI storage classes
SSD and SATA classes, ext4 or xfs, raw block mode, topology awareness and volume snapshots.
Cloud controller manager
A Service of type LoadBalancer becomes a VPSie load balancer, with health checks and private VPC options set by annotations.
Cluster API provider
Declarative, GitOps-style cluster lifecycle through the Kubernetes Cluster API.
Preview add-ons are available to early adopters on request and are not covered by the VPSie SLA.
Learn more: Install K9s Container registry Docker one-click
What teams run on it
Microservices and SaaS
Rolling deploys, self-healing pods and horizontal scaling.
CI/CD and GitOps targets
Argo CD, Flux, GitLab or GitHub pipelines pushing to a cluster you control.
Stateful workloads
Databases, queues and caches on block-storage volumes.
Dev, staging, production
Separate clusters per project, each sized on its own.
Questions about Kubernetes
Which Kubernetes do you run?
Upstream Kubernetes bootstrapped with kubeadm, not a fork. Versions in the console catalog are listed when you create a cluster.
Is the control plane highly available?
With more than one master, yes. The API endpoint sits behind an HAProxy load balancer and etcd survives the loss of a master.
Can I automate it?
Yes. The API, the Go SDK and Terraform all create and scale clusters and node groups.
Does it autoscale?
You can scale masters and workers on demand today. The node autoscaler is in preview; ask us to join.
Where is the cluster network?
Inside a VPC you own. Nodes use private addresses, and the API load balancer is the one entry point to the control plane.
Related VPSie pages
Cloud servers Platform services Cloud networking Cloud storage VPSie public cloud
Create a cluster
See the live plan and node price in the console before you commit.