I do a lot of puppet development and parts of my development cycle require me to spin up a whole virtualization cluster with a master node and at least one work

I do a lot of puppet development and parts of my development cycle require me to spin up a whole virtualization cluster with a master node and at least one worker node ot test interactions - possibly even starting vms in that cluster. I can certainly start simulating that with docker, but I'd prefer to be as close as possible. This does indeed require substantial RAM, but doesn't require any ECC or stable uptime. I can certainly do that part of my development on some bit of dedicated hardware that sits in the office, but that means I need networking there, which means I can't code on the road etc.