I think OP was referring to things like vim configs, shell customizations, nginx configs, etc. There's lots of computer customizations I could think of that aren't always thought about in context of storing in a git repo. Also you could have large files such as market data, movies, or music collection that aren't a good fit for a git repo.
I noticed the usage of computers changed drastically over years. When I started last millennium it was normal for people to have large amount of installed tools with large amount of customizations.
I have a bunch of custom-compiled software with options for my specific needs, and my two main computers (PC and a laptop) have both huge history of various workaround that I don't even remember any longer and would have to research all of them again. I also have a bunch of Virtualbox VMs. These are a Windows VM configured for remote work, couple Linux VMS with configurations for embedded development (I do ARM development). I have one dedicated to Matlab, Simulink and LabView and host of plugins and customizations.
While keeping dotfiles and scripts is easy (I just have a repository on bitbucket) I would still have to spend a day or two just getting everything in order and probably much more to get these VMS for embedded development which is just full of workarounds for various problems.
As a contractor paid by the hour I am aware of my hourly rate and this helps me put things in perspective. Once you calculate the value of services and tools what seemed expensive frequently becomes dirt cheap when you factor time saved or risk averted.
For backup I keep USB docking station and two 4TB HDDs. I try to keep one of them at a different site. I also keep a copy on an online service mostly for convenience but also as another layer of protection.
