Check your inbox for a confirmation email where you can adjust your preferences and even join additional groups. KubeVirt was founded by Red Hat in 2016 around the idea that Kubernetes could run virtual machines alongside containers . We wanted to allow users to run legacy workloads on the same platform as their modernized applications. That is still the job, but since v1. 0 other projects have demonstrated that they can build on it and integrate KubeVirt in a broader ecosystem. Let's discuss some of what the next 10 years of KubeVirt will bring.

Other projects such as Open Sovereign AI Cloud (OSAC) use it for VM, cluster, and bare-metal services. Cluster API Provider KubeVirt (CAPK) uses KubeVirt VMs as Kubernetes nodes. Flight Control runs a virt-launcher on the edge without the full KubeVirt control plane. Today KubeVirt is a CNCF incubating project, with contributors from Nvidia, Microsoft, AMD, Intel, SUSE, and others. The community is working toward KubeVirt's graduation as a CNCF project. The enhancement trackers for v1. 10 show what "next" means in practice. 10 the queue went from a small set of items to more than a hundred.

9 included dozens of new proposals. The KubeVirt Enhancement Proposal (VEP) process is how we keep the core stable while that queue grows: designs before code, one tracker, and ownership by special interest groups (SIGs). In this article, I will share some of the work that will shape the next releases: devices and new hardware, extending KubeVirt without growing core, confidential VMs, the VM as a sandbox, and pluggable virtualization stacks. My fellow KubeVirt contributors, Fan Zhang and Harshit Gupta, help explain the features they are working on.

GPUs on VMs can already be claimed through Kubernetes Dynamic Resource Allocation instead of using device-plugins. This functionality is beta in v1. The same claim model is being extended to generic host devices, network devices such as SR-IOV, and to CPUs. In the near future, a VM will be able to ask for a GPU, NIC, and CPUs and have Kubernetes allocate them as a coordinated set. There is also work to allow KubeVirt to build those claims from the VM spec and an input from vendor policy, instead of asking every user to write ResourceClaim objects by hand.