My homelab started the way most do: an old desktop that was about to be thrown out got a fresh install of Proxmox VE instead. One node, a handful of VMs, no plan.

A year later it’s a small cluster, and here’s roughly how it grew:

• Node 1 — the original desktop, now mostly running the “boring but important” stuff: DNS (Pi-hole), a reverse proxy, and backups.
• Node 2 — a mini-PC bought specifically for the homelab, running the heavier self-hosted apps.
• A managed switch with VLANs, because everything living on one flat network was the first mistake I fixed.

Things I got wrong early on

No backups until something broke. Proxmox Backup Server takes maybe 20 minutes to set up and I put it off for months. The first time a VM’s disk corrupted, I understood why that was a bad idea.

One giant VLAN. IoT devices, work-adjacent VMs, and “trusted” management traffic were all on the same subnet. Splitting into VLANs (management, trusted, IoT/guest) with firewall rules between them was a weekend project that paid for itself immediately.

Snowflake configs. Every service was set up by hand, once, and never documented. I’m slowly moving everything to Docker Compose files and Ansible playbooks checked into a private git repo, so a node dying doesn’t mean starting from scratch.

What’s running today

A short list, roughly by usefulness to me day-to-day: Pi-hole, a reverse proxy (Caddy) with automatic TLS, Immich for photo backups, a couple of monitoring dashboards, and — fittingly — this very blog, once I get around to self-hosting it.

If you’re starting out: buy less hardware than you think you need, and spend the time you save on backups and network segmentation instead. Both are much more painful to add after the fact than before.