Skip to main content
Version: main 🚧

Start here

vCluster provisions isolated clusters, delivering managed Kubernetes, Slurm, inference, Ray, and other cluster types. The clusters can go on your existing infrastructure, or they can be built from scratch on bare metal and VMs. Each tenant gets a full Kubernetes API experience while the virtualized control plane stays completely invisible to them.

Try it on your infrastructure​

Planning a customer-facing or production platform?

These quick starts prove that vCluster runs on the infrastructure you have. They don't choose a production architecture. Once a quick start works, read Security and tenant isolation and decide whether tenants share hardware. Then work through Plan your cluster.

Match your situation to a guide​

  • Plan a complete service for your tenants, beyond a running cluster → Bare metal, VMs, Kubernetes, AI platforms, or Team and CI clusters
  • Build a production Kubernetes platform from bare metal or VMs → Standalone
  • Use an existing cluster for a GPU cloud, AI inference service, or regulated platform that needs dedicated infrastructure per tenant → Private Nodes
  • Use an existing Kubernetes cluster for internal platforms, CI, testing, or development with trusted teams → Shared Nodes
  • Create a Kubernetes cluster on Docker for local development or CI → Docker (vind)

Still unsure? Read What is vCluster? to understand the deployment models and tradeoffs.