KubeVirt integration in OpenShift and Rancher
Best of Both Worlds
Traditional enterprise virtualization clusters run on physical servers, on which Kubernetes clusters, whose nodes run on virtual machines (VMs), are then based. This kind of architecture is fraught with many drawbacks. On the one hand, two layers means additional license and subscription costs. On the other hand, both the VM and container layers virtualize resources such as networks and storage, which impairs performance in many installations.
In some organizations, Kubernetes nodes already make up the majority of VMs in the cluster while the number of classic VM-based applications steadily decrease. The tide has turned from many VMs and a few containers to many containers and a few VMs. Kubernetes can take care of these remaining VMs with an add-on named KubeVirt, which allows you to remove the obsolete virtualization layer from the data center and place the server hardware with all the containers and VMs entirely under Kubernetes' control.
Add-Ons for VMs
Although you usually think of containers when you hear the word Kubernetes, strictly speaking, Kubernetes is an open cluster management framework. Given the right choice of custom resource definitions (CRDs) and add-ons, the framework is not only capable of managing containers but can also handle virtual networks, distributed storage, and, of course, the matching security rules and access authorizations.
Extensions for Kubernetes save their current configurations in the system's own key-value store (typically etcd
) and provide a REST API. They use these configurations to field instructions from the Kubernetes framework. To use the add-on features, you need to formulate a matching request to the Kubernetes framework. The "child" used in the request (i.e., the resource type to be managed) then points to the add-on API to which Kubernetes transfers the request.
Extensions also use the Kubernetes
...Buy this article as PDF
(incl. VAT)