I think Kubernetes is an incredible piece of software, but by the time you've really adopted it in the way the ecosystem recommends you have dozens and dozens of moving parts just to run a single service on it. Conversely, the FreeBSD ecosystem would push you to more OS-based solutions (such as jails and the backup solutions mentioned in the article).
Could you do the same approach on Linux? Sure, and I did that for a decade, but these days if you are googling around for best practice that's not remotely where you're pushed, or where the upstream effort is being spent.