A funny thing happened today: after a meeting I ended up on the phone with a former colleague and we drifted into the ongoing “is it really jails?” naming discussion around my NetBSD experiment.
He pointed me to the FreeBSD Handbook and suggested I look again at how jails are actually described there. That sent me down a small rabbit hole. The more I read, the less clear-cut the distinction felt.
At the lowest level, FreeBSD jails are essentially a kernel mechanism that attaches an identity to processes and restricts visibility and interaction. Many things people associate with “modern jails” today - VNET networking, ZFS-based setups, orchestration frameworks - often live a layer above that core mechanism in tools like BastilleBSD and similar projects.
Interestingly, FreeBSD’s own docs sometimes describe jails as the subsystem that enables containers, and the industry term “container” shows up quite regularly there as well. FreeBSD can even run OCI containers via the Linux compatibility layer. Which made me wonder: have “jails” gradually become something like a brand name for the FreeBSD flavor of containers in people’s minds?
I’m honestly still undecided. The more I read, the more it feels like the answer depends a lot on the perspective and background one brings to these terms.
Curious what the poll will say — please vote if you haven’t yet. And if the right name isn’t listed, feel free to drop suggestions in the comments 🙂
#netbsd #freebsd #openbsd #dragonfly