Running a company on hardware we own

Most of Kitsune's internal tools (CRM, contract signing, single sign-on, DNS, file storage, monitoring, AI services) run on a handful of Proxmox hosts we own and operate. It's not ideology. For steady, predictable workloads, owning the hardware is cheaper, more controllable, and free of vendor surprises. The catch is that you also own the operational work, and that's the part most people skip.

One container, one job

Every service is its own unprivileged LXC container with a dedicated data directory and a single responsibility. That keeps blast radius small and makes each service easy to reason about, back up, and restore on its own.

One door in

A single Authentik instance handles login for everything internal, enforced at the reverse proxy through Caddy forward-auth. Public traffic terminates TLS with automatic wildcard certificates, and anything exposed to the internet has intrusion protection in front of it. New service? Add a proxy route, add an auth rule, done.

The part that makes it real

Application data lives on ZFS datasets with scheduled snapshots and off-box backups, and we've actually tested a restore, not just assumed one would work. Recovery paths are written down. That's the line between a homelab and infrastructure: a homelab is impressive until it breaks; infrastructure is something you can confidently put back together.

Why it matters to clients

When we offer infrastructure work, it isn't theoretical. We run the thing we sell, we've debugged it at 1am, and we know exactly what "operable" costs, because we pay it ourselves.