Sure, ZFS snapshots are dead simple and fast. But you’d need to ensure that each container and its volumes are created in each respective dataset.

And none of this is implying that it’s hard. The top comment was criticizing OP for using VMs instead of containers. Neither one is better than the other for all use cases.

I have a ton of VMs for various use cases, and some of those VMs are container/Docker hosts. Each tool where it works best.