KVM vs OpenVZ vs LXC — Why the Virtualisation Type Matters

Own kernel, honest RAM, real swap, BSD or Windows images: full virtualization changes what a VPS can run. Here is why every plan here is KVM.

O texto longo do artigo está em inglês. Cabeçalho, catálogo e checkout estão traduzidos.

Containers vs virtual machines

OpenVZ and LXC put every tenant on the host's shared kernel; memory is a burst-accounted pool that the host reclaims whenever a neighbour spikes. KVM instead hands each server a real kernel of your choosing with reserved RAM and its own swap. The monthly spread between the two models is a few dollars at catalog prices — Sentry sits at $7.50/mo list — and the behavioural gap opens precisely when your workload stops being a demo. Every plan in this catalog is KVM for that reason.

What KVM gives that containers cannot

• A kernel you pick: custom builds, BPF tooling, out-of-tree modules, DKMS drivers, WireGuard without provider blessing.
• Windows and BSD images, which container-only shops structurally cannot sell.
• Real swap, so a memory spike degrades latency instead of triggering a hard kill.
• Unrestricted Docker and Compose, nested virtualization on the VPS lines, raw sockets for packet tools.
• RAM that is yours: the 4 GB on Sentry is 4 GB of reserved ECC, not a slice of a contended pool.

The failure modes of oversold containers

Ask anyone who ran a busy site on oversold OpenVZ: the host hits memory pressure, the burst credit vanishes, and the OOM killer picks your database despite “your” usage looking small. Sometimes the kernel lacks a feature your stack needs, or an iptables module is simply absent, and no ticket changes either. Those are architectural outcomes, not one bad provider. Full virtualization removes the entire category, which matters more as workloads grow from toy to production.