Published 23 minutes ago

Ayush Pande is a PC hardware and gaming writer. When he’s not working on a new article, you can find him with his head stuck inside a PC or tinkering with a server operating system. Besides computing, his interests include spending hours in long RPGs, yelling at his friends in co-op games, and practicing guitar.

Sign in to your XDA account

Over the years, I’ve tinkered with all sorts of home lab paraphernalia, including repurposed PCs, efficient mini-PCs, and even enterprise-grade hardware. But despite owning significantly more powerful devices, I keep coming back to my Raspberry Pi collection for low-power projects. As a result, I’ve got a couple of Single-Board Computers that I tend to use for an experiment or two, before setting them aside and forgetting about them for the next couple of weeks while I work with containers and VMs.

So, I figured I could try integrating my SBC collection into my home lab. After all, I’m already using my primary Raspberry Pi 5 system to self-host essential containers and maintenance tools, and adding more boards to the mix should help me create reliable clusters for my DIY experiments. And turns out, Raspberry Pi systems work really well in cluster setups.

Related

5 cool things I learned from my first Raspberry Pi

Despite its unassuming appearance, the Raspberry Pi can teach you a lot about computing

I initially went with a Kubernetes cluster

My RPi boards were able to handle this beast of a platform

As one of the most essential container orchestration tools, Kubernetes was a familiar term even before I began my descent into the home lab rabbit hole. I always thought I’d need hardcore server systems to tinker with its auto-scaling prowess and automation tools. But that couldn’t have been further from the truth.

K8s is a lot more lightweight than I expected, and the same holds true for containers. In fact, a single Raspberry Pi system can deliver surprisingly decent results when paired with this platform. Throw in a couple of these tiny boards, and you’d have a cluster of Kubernetes nodes ready to do your bidding.

I’ve even used a non-Raspberry Pi board in my K8s cluster when I built it for the first time, and it worked pretty well alongside my green tinkering companions. Sure, I was limited to Arm images instead of their AMD64 counterparts, but Kubernetes’ auto-deployment, self-healing, and management tools worked without any performance or compatibility issues. I used this setup sometime ago for my uni projects, since VM-based Kubernetes nodes wouldn’t help me get accustomed to the quirks of a bare-metal cluster.

But Docker Swarm is the better option for the average tinkerer

I went down the Kubernetes route to train my DevOps skills, but Docker Swarm works just as well for home labbers looking for the high-availability benefits of a cluster without the extra complexity of typical K8s setups. Deploying Docker Swarm on RPi boards is extremely simple, even if you’re not on Raspberry Pi OS – and I say that as someone who went with a CLI distro when I tried turning my SBC collection into a Swarm configuration. It also doesn’t have the same complexity as a Kubernetes setup, since I could run containers as stacks instead of rewriting config files to suit a K8s environment.

I used Portainer as the management interface, as Docker Swarm doesn’t include a web UI I could use to oversee my container collection. But it’s so light that it doesn’t impact the Raspberry Pi manager nodes. Personally, I would’ve kept this setup in the long run, if not for the fact that I already have some high-availability nodes in my home lab.

Canonical’s MicroCloud has become my favorite tool for reviving spare SBCs

It turned my Raspberry Pi systems into a proper home lab

My K8s and Docker Swarm configurations were good enough for containers, but virtual machines are a different story. Sadly, most of the popular virtualization platforms favor x86 CPUs and have yet to be ported over to the Arm space. I guess I could’ve given a Proxmox-based Raspberry Pi cluster a shot, but I’ve already dealt with the deal-breaking issues of deploying a community port of PVE on my RPi 5.

But when I was messing around with Canonical’s MicroCloud, I thought of installing it on a Raspberry Pi. While it didn’t work so well with Raspberry Pi OS, Ubuntu Server was a different story. MicroCloud not only spun up on my tiny tinkering boards, but I was even able to link them together via MicroCeph.

The performance was nothing to scoff at, either. My main control node was able to handle a few VMs (albeit their CLI variants), alongside a couple of LXD-powered containers. Since the rest of the cluster nodes are weaker than my Raspberry Pi 5, I use them for container workloads and the occasional virtual machine.

But you can also use standalone RPis for cool projects

If you’re not a fan of clustering, there are plenty of other fun ways to put your spare Raspberry Pi boards to good use. While these tiny systems aren’t ideal for hardcore NAS workloads, they can double as solid file-sharing servers or even secondary, remote-based backup hubs. Alternatively, if you’ve got a Raspberry Pi AI kit lying around, you can slap it on top of an RPi 5 and use the SBC as a powerful Frigate NVR that’s as good at recording footage as it is at motion detection. Or, you can even turn your adorable little tinkering partner into a locally-hosted Office 365 alternative with Nextcloud and a couple of companion apps.

Related

I use these 5 tools to turn my Raspberry Pi into a self-hosting behemoth

Even a Raspberry Pi can become a containerization war machine with the right set of apps