Yes.

I also have a bunch of other VMs. For example, I keep my toolchains for embedded development in separate VMs so that I can trust they still work years later when I need to do some quick correction to some project of mine. These have auto updates turned off. I hate when I want to quickly change something and program the device but first I have to investigate why the tool chain no longer works.

Not all of these VMs are active at the same time. I usually have 2 or 3 VMs active, some paused and most turned off.

I do similar things. Everything I run is built on top of LXD or KVM so that it remains compatible across time and is not beholden to anything but the Linux kernel, with disk images backed up at key points of the installation process. All of the build scripts are housed at [1] and then I have a set of machine-specific scripts that leverage those to rebuild any of my virtual systems in a single command with no data loss. Press a button, walk away, and in half an hour I have a pristine system with everything I need installed and my data drive mounted.

Any time I'm working on a machine that was set up non-deterministically, I feel vulnerable...

[1] https://github.com/kstenerud/virtual-builders

This brings to mind, fondly, my gentoo days. Just never had time to continue to spend that much time controlling the config vs using it.

This is a great idea, thank you!

I recently started playing around with libvirt/QEMU on Linux and was amazed how well the VMs work, especially when you have two video cards plugged in and can do PCI passthrough.

It's so seamless. I now run my Windows "VM" on that, and works fantastically. I have it installed to a separate SSD as well (I was curious if it would work) and performance is excellent.

Yeah I was watching a video[1] about QEMU vs virtual box comparison yesterday on youtube and the guy doing it found that the QEMU version was faster than his host machine in some cases. It was confusing!

[1] https://www.youtube.com/watch?v=hOKQfeUNTCE