From Pre-Population To Disasters: Manage And Protect the State Of VMs

KubeVirt makes it possible to run traditional Virtual Machine workloads in a Kubernetes cluster. Since VMs are typically stateful and not as homogeneous as containerized applications, additional care must be given to ensure that VM state is properly initialized, managed, and protected. We will explore how the KubeVirt storage layer bridges the gap between QEMU/KVM Virtual Machines and K8s storage primitives to provide a feature-rich API that manages data for the entire lifecycle of a VM. We will discuss how new VMs can be created with pre-populated disks based on “golden images” and how running Virtual Machines can be safely snapshotted/restored. Disaster Recovery workflows are enabled by the VirtualMachineExport API as well as integration with Velero. Future initiatives, such as Volume Populator support will also be discussed. You will come away with enough of a high level understanding of the KubeVirt storage APIs and architecture to make meaningful contributions.