Your Pods can be stateless, your VMs cannot! How to manage state in k8s
summary
Shows that Kubernetes can run VMs as processes using KVM/QEMU and a VM API (VirtualMachine and VirtualMachineInstance CRDs), but storage is the blocker: a customer test that scales PVCs from dozens to thousands of volumes breaks classical arrays during heavy metadata ops. Describes using Portworx data platform to provide scalable, platform‑agnostic CSI storage, support persistent volume claims for VM disks, enable live migration and namespace-level async DR replication across clusters — and reports the approach works after fixing backend limits.
AI-written summary · about
chapters (13)
- 0:00Introduction and audience questions
- 1:22Storage in Kubernetes basics
- 5:15Challenges with CSI storage
- 9:09Container-native storage approach
- 10:59Why run VMs in Kubernetes
- 13:05Technical deep dive on VMs in Kubernetes
- 17:21Kubevirt architecture overview
- 20:07Storage for virtual machines
- 23:15Requirements for VM storage
- 25:05Portworx data platform overview
- 26:30Disaster recovery with Kubernetes
- 28:58Demo and hands-on labs invitation
- 29:42Q&A session
related talks
28mFrom Pre-Population To Disasters: Manage And Protect the State Of VMs
29mStateful Workloads in Kubernetes: A Deep Dive
37mKubernetes Storage 101
30mKubernetes Storage Cheat Sheet for VM Administrators
20mKubernetes for the Virtualization Admin
34mData Management in Kubernetes – Backup, DR, HA
19mKeynote: Debunking the Myth: Kubernetes Storage is Hard
49mDevOps: Kubernetes Storage - Myths, Facts and Tips