News
BackManaging Entire Clusters Using K8s: Cluster API
Kubernetes is the industry standard for deploying containerized workloads such as pods, persistent volumes, and other components, as well as for their ongoing management. But what about the cluster itself? Managing a cluster is complex. This is exactly where the Cluster API (CAPI) comes in: it allows you to describe entire clusters declaratively. This automates the entire management process – from setup and updates to deletion – and makes it fully reproducible.
Declaring K8s clusters at cloudscale in YAML
Thanks to the Cluster API (CAPI), entire clusters can be completely described using YAML files and managed automatically. To this end, CAPI introduces a number of additional architectural components that allow heterogeneous cloud APIs to be hidden behind a common abstraction layer.
The "Cluster API Provider for cloudscale.ch" (CAPCS) serves as the link that enables a fully functional Kubernetes cluster to be created on our infrastructure from a simple YAML configuration. CAPCS allows you to specify, for example, which compute flavor your control plane and worker nodes should be assigned, and acts as an interface between Kubernetes' internal capabilities and the cloudscale API.
Fewer errors, greater efficiency
By managing your clusters using CAPI, you ensure that each cluster precisely matches its specifications and largely eliminate potential errors caused by manual actions. The infrastructure for your workloads is predictable and reproducible; variations between environments (e.g. development, testing, staging, production) or teams are minimized. Even a temporary test cluster can be set up quickly and exactly according to your standards – and can be deleted just as quickly afterwards.
However, CAPI does not just support you with short-lived clusters – on the contrary: when it is time to upgrade the Kubernetes version, your nodes are updated one after another in a "rolling upgrade". This reduces the risk associated with an upgrade and minimizes the resulting downtime. And even between planned changes, continuous reconciliation ensures that your infrastructure always matches its definition.
Try it out straight away
CAPCS is available on GitHub and complements the existing Kubernetes integrations (CSI and CCM) with cloudscale. In addition to the CAPCS code, you will also find a concise "Getting Started Guide" on GitHub.
Our engineering blog goes into even more detail: Starting from a "seed cluster", Michael walks you through, step by step, how to initialize a management cluster, create your first workload cluster, and perform a Kubernetes upgrade – complete with configuration snippets, specific commands, and plenty of links to further resources.
Many of our customers are already using Kubernetes to manage their workloads. With the Cluster API Provider for cloudscale.ch (CAPCS), you can now manage entire Kubernetes clusters directly from within Kubernetes. Whether for quick tests or consistent "day 2 operations": Specify entire clusters in YAML files and let Kubernetes take care that the right infrastructure is provisioned at all times.
We solemnly declare... clusters!
Your cloudscale team