KubeVirt: What If Kubernetes Orchestrated Your VMs?
summary
KubeVirt runs full virtual machines by launching QEMU inside Kubernetes pods (a virt-launcher / QEMU-in-pods model), reusing Kubernetes storage and networking while adding features like live migration, vGPU, NUMA and confidential compute. VM image workflows are either embedding large QCOW2 files in containers (slow) or using CSI image import plus PVC cloning (faster) — initial imports can take 10–15 minutes unless cloned. Networking requires Multus/OVN-Kubernetes and IPAM for persistent IPs (otherwise DHCP/IP persistence is problematic), and KubeVirt is suited for running a few VMs in container-first clusters rather than replacing OpenStack/VMware.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 2:08Containerization vs virtualization
- 4:36Kubevirt principles
- 7:25Kubevirt architecture
- 9:30VM vs containers differences
- 11:16Image storage issues
- 13:44Image management solutions
- 17:42Network issues with Kubevirt
- 22:18Summary and recommendations
- 23:24Use cases and alternatives
- 24:07Q&A
related talks
43mKubeVirt - Kubernetes, Virtualization and Your Future Data Center
32mKubeVirt - making running virtual machines in a k8s cluster
25mKubeVirt: Converge IT infrastructure into one single Kubernetes platform
38mKubeVirt Intro: Virtual Machine Management on Kubernetes
30mKubeVirt: bringing virtual machines in a Kubernetes world
43mMaking Running Virtual Machines in a Kubernetes Cluster a Mainstream Activity
31mKubeVirt: What if? What if We Could Create Virtual Machines in Our K8s Cluster?
42mVirtual Machine Orchestration in Kubernetes Utilizing KubeVirt