Kubernetes Bytes Kubernetes for VMware Admins: Understanding KubeVirt
Mar 6, 2026
Janakiram MSV, architect and analyst focused on Kubernetes and cloud-native tech, explains KubeVirt and why VMware admins should care. He maps Kubernetes concepts to vCenter. He breaks down KubeVirt architecture, how VMs run as pods, storage and networking choices, scheduling differences, migration limits, and practical testing and skills for admins moving to Kubernetes.
AI Snips
Chapters
Transcript
Episode notes
How KubeVirt Runs VMs Inside Pods
- KubeVirt wraps a KVM VM process inside a pod so Kubernetes sees a pod while the guest sees a VM; key components are the API server, controller, and a node-side handler daemon.
- Janakiram highlights CRDs for VirtualMachine and VirtualMachineInstance separation, enabling declarative desired state and runtime instances for graceful restarts and live migration.
Declare VMs With YAML And Use Kubernetes Primitives
- Define VMs as CRD YAML objects and apply them just like other Kubernetes resources to create VirtualMachine and VirtualMachineInstance objects.
- Use existing Kubernetes primitives for storage (PersistentVolumes) and networking (Services) while relying on CDI and Multus for VM disks and multi-NICs.
Use Multus For Networks And CDI For VM Disks
- For VM networking use Multus to attach multiple CNIs and replicate multi-NIC enterprise patterns; for VM storage use CDI to import disk images into PVCs.
- CDI ingests images from URLs, registries, or PVCs and produces the persistent volumes that act like VMDKs for VMs on Kubernetes.