OP here - to me, it is not. For most services I picked SQLite, so the whole backup is one file / directory. Software can easily be updated or test instances st

OP here - to me, it is not. For most services I picked SQLite, so the whole backup is one file / directory. Software can easily be updated or test instances started since all is driven by docker. For the few services that need a DB there is a central one serving all hosts. Backups regularly (on a nightly basis where sufficient) go to the NAS. The VM docker is running on gets regular snapshots. Some initial setup and that's it. The whole setup is as movable and reproducible as I could justify making it over a weekend. Securing theres is an interesting point. I take the risk of a local attacker. Other than that all I got to keep secure is the VPN, which I would have to do anyway. I am not trying to make it sound like this is something you spend 10 minutes on and call it a day. I am doing this (running prod services) to various degrees for nearly two decades. If you are just getting into all of this, sure, it will take more time and dedication. I see your point, but I do not necessarily agree with it. I would argue that even with "nothing controversial or interesting" it still is a loss of control. You do not control your data. You do not control if the service will stay available to you and / or your friends.