Keynote: Our Journey Towards KubeVirt Based HyperConverged Infrastructure...
summary
Migration moves a Docker-based internal cloud of ~8k containers on 700+ baremetals and ~200TB memory onto a KubeVirt HCO stack to run VMs and pods side-by-side. Network and DNS get solved by a forked kube-secondary-dns plus a controller that creates records for VMIs and pod ports; IPAM and secondary interfaces use multiple CNIs; provisioning drops from minutes to seconds using Longhorn backing image snapshots; live migration uses RWX root disks; quota and scheduling rely on Apache YuniKorn and workflow automation via StackStorm.
AI-written summary · about
chapters (9)
- 0:00Container era introduction
- 2:09VMs still matter
- 4:35Container Cloud challenges
- 6:47KubeVirt architecture
- 8:16Networking for stable IP/DNS
- 10:04Fast VM provisioning
- 11:08Live migration for stateful apps
- 11:30Quota for mixed workloads
- 12:14Migration journey
related talks
31mElephant in the room
28mHow We Tackle KubeVirt’s Growth and Scalability
18mKubeVirt Alchemy: Transform VMs into K8s Workloads
32mKubeVirt - making running virtual machines in a k8s cluster
43mMaking Running Virtual Machines in a Kubernetes Cluster a Mainstream Activity
42mVirtual Machine Orchestration in Kubernetes Utilizing KubeVirt
43mKubeVirt - Kubernetes, Virtualization and Your Future Data Center
04mLightning Talk: Kubespray CI with KubeVirt on Baremetal