KubeVirt — Run Virtual Machines on Kubernetes

What it is

KubeVirt is a Kubernetes extension that adds virtual machine workloads as first-class citizens to your cluster. Instead of choosing between containers and VMs, KubeVirt lets you run both side by side, sharing the same networking, storage, and scheduling infrastructure.

The project targets platform engineers managing hybrid workloads where some applications cannot be containerized yet. Legacy Windows apps, licensed software tied to specific OS images, and workloads requiring full kernel access are common use cases.

How it saves time or tokens

Without KubeVirt, teams maintain separate infrastructure stacks for containers and VMs, doubling operational overhead. KubeVirt eliminates this split by letting you define VMs with the same kubectl commands and YAML manifests you already use for pods. One API, one control plane, one set of RBAC policies. Migration from standalone VMware or KVM setups becomes incremental rather than a big-bang rewrite.

How to use

1. Install the KubeVirt operator and custom resource on your cluster:

export VERSION=v1.3.0
kubectl apply -f https://github.com/kubevirt/kubevirt/releases/download/${VERSION}/kubevirt-operator.yaml
kubectl apply -f https://github.com/kubevirt/kubevirt/releases/download/${VERSION}/kubevirt-cr.yaml

1. Create a VirtualMachine manifest and apply it:

apiVersion: kubevirt.io/v1
kind: VirtualMachine
metadata:
  name: demo
spec:
  running: true
  template:
    spec:
      domain:
        devices:
          disks:
            - name: disk0
              disk:
                bus: virtio
        resources:
          requests:
            memory: 512Mi

1. Access the VM console with virtctl console demo.

Example

# List running VMs
kubectl get vmi

# SSH into a VM via virtctl
virtctl ssh user@demo

# Live migrate a VM to another node
virtctl migrate demo

# Stop a VM
virtctl stop demo

Related on TokRepo

• AI Tools for DevOps — more infrastructure automation tools for Kubernetes environments
• Automation Tools — scripts and workflows for infrastructure management

Common pitfalls

• KubeVirt requires hardware virtualization support (Intel VT-x or AMD-V) on worker nodes; nested virtualization in cloud VMs adds latency
• VM images stored as PVCs can be large; plan storage capacity and use CDI (Containerized Data Importer) for efficient image management
• Network policies that work for pods may not cover VM traffic correctly; test VM-to-pod connectivity explicitly after setup