I'm learning Kubernetes and want a hands-on lab on my local Windows laptop instead of only studying the theory. I currently use WSL with Ubuntu and would like to create an environment where I can deploy applications, manage the infrastructure, troubleshoot failures, and deliberately break and rebuild things. What setup would be a good starting point, and when should I move from a local single-machine cluster to multiple virtual machines or physical nodes?
4 Answers
A larger VM-based lab could use one controller and three workers, or three controllers and three workers if the hardware can handle it. Three controllers are useful for practicing control-plane failover, but they are not necessary for an introductory lab. Proxmox is worth learning later because templates make it easier to clone and scale nodes; kubeadm can then be used to bootstrap the cluster.
For getting started on Windows, KIND is probably the easiest option. It runs Kubernetes nodes in containers, starts quickly, and can be deleted and recreated whenever you need to reset the lab. Once you understand the basics and need more realistic node-level practice, you can move to Linux virtual machines or a hosted cluster.
WSL may become awkward when you try to run several independent nodes. A practical progression is to begin with KIND, then try k3s on a few Linux VMs. That gives you a lighter cluster while still letting you practice services, networking, storage, upgrades, and recovery in a more realistic environment.
If you have spare hardware, install Proxmox and create three Ubuntu Server guests: one control plane and two workers. It is not highly available, but it is a solid lab for learning cluster administration. You could also use Hyper-V or VirtualBox on the laptop, as long as the machine has enough memory and CPU and the VMs share a private virtual network.
The laptop has 8 CPU cores and 15 GB of RAM, so I’ll probably start with a smaller setup rather than trying to run several controllers.