One thing that irks me about these solutions is that they seem to scan my folders each time they want to backup. Are there tools that are smarter about this? For e.g., while running, they could keep a log of what's changing and only scan those while backing up.
I may have looked at Duplicacy. I'm sure that I looked at one of Duplicacy [1], Duplicati [2], and Duplicity [3]. Whichever one that was, I kept getting it mixed up with the other two when looking for information online, and finally said "screw this" and bought Arq.
I've used it on my macs for years without any issues at all. I switched after Time Machine broke down for the n'th time in a month saying it needed to recreate the backup, and not once in the 3-5 years i've been using it has it every given me any problems with broken repositories, and every integrity check/restore has succeeded.
Arq on Windows is a different beast though. I'm sure it's technically solid, but the UI leaves a lot to be desired. On windows boxes i default to Duplicaty.
Did you mean Duplicacy or Duplicati?
On my servers i use Borg like any sane person would, but the lack of a good client UI makes scheduling backups on a personal computer a lot more work than i'm willing to put in.
I've found arq backup overall slow once you start hitting 0.5TB overall. It's a design issue.
I believe most operating systems have a mechanism for this. E.g. on macOS it's called fsevents, Linux has inotify