Build your first homelab

A homelab is a small, always-on machine you use to run things - a hypervisor, a few virtual machines, a stack of containers - so you can host your own services and learn how real infrastructure behaves. The trap is treating it as a hardware collection. The point is what runs on it. Here is a starting point that stays cheap and grows with you.

1. Start with one box, not a rack

Almost everyone's first homelab should be a single, quiet, low-power machine. A mini PC with a modern CPU, room for 32GB or more of RAM, and a spare drive slot will run more than a beginner thinks. It draws a few watts idle, makes no noise, and costs less than the "deal" on a used enterprise server that will heat your office and spin like a jet.

Buy for RAM and a couple of network ports before raw CPU. Virtual machines and containers eat memory long before they exhaust cores, and a second NIC opens up networking projects later.

2. Pick a hypervisor

The hypervisor is the layer that lets one box pretend to be many. The standard free choice is Proxmox VE: it runs virtual machines and Linux containers, has a clean web UI, does backups and snapshots, and has a large community. Install it on the mini PC and you have the foundation for everything else.

If you would rather live entirely in containers, you can run Docker on a plain Linux install instead. Proxmox is the more flexible base and the one most guides assume, so it is the safe first pick.

3. Give it a real network

This is where a homelab stops being a single computer and starts being infrastructure. You do not need much to begin, but a little intent pays off:

• A managed or 2.5-gigabit switch so you can segment traffic and move data quickly between the lab and your NAS.
• A router you control - OpenWrt, OPNsense, or similar - if you want VLANs to isolate lab experiments from the devices your family depends on.

Both are covered at HomeSwitches. Isolating the lab on its own VLAN early is the difference between a broken experiment being a shrug and being a household outage.

4. Add storage deliberately

The mini PC's internal drive is fine for the hypervisor and a few VMs. The moment you host media, backups, or shared files, point the lab at a proper NAS instead of stuffing everything into the one box. Keeping compute and storage separate means you can rebuild or replace the lab without touching your data. Pick the drives for that NAS at NASdisks, CMR only.

5. Run something useful on week one

Hardware sitting idle teaches you nothing. Give the lab a job you will actually notice if it breaks - that is what makes the learning stick. Common, high-value first services:

• Pi-hole or AdGuard Home - network-wide ad blocking, an instant "everyone in the house benefits" win.
• A reverse proxy (Nginx Proxy Manager, Traefik, or Caddy) so your services get clean names and TLS.
• Home Assistant if you are heading toward smart-home control - the local-first automation hub that ties devices together without the cloud. The devices that work well with it are catalogued at Domotics.
• A backup target for your other machines.

Pick one, get it working end to end, then add the next. Breadth comes from repetition, not from spinning up ten half-finished stacks.

6. Back up the lab itself

A homelab accumulates configuration you will not want to rebuild from memory. Turn on Proxmox's scheduled backups, send them to your NAS, and test a restore once. The same 3-2-1 rule that protects your files protects your lab: three copies, two media, one off-site.

The short version

One quiet mini PC, Proxmox on top, a network you control with the lab on its own VLAN, storage kept separate on a real NAS, and one genuinely useful service running by the end of the first week. That is a homelab that teaches you something and survives contact with a real household. Start with the box at ServerRigs and the network at HomeSwitches.