The Backup Paradigm
Over the years, I've become convinced, that 'backups' are a bad paradigm, and should be left to edge-cases. We should instead focus on 'synchronization' where possible.
Where backups say 'make a copy, on tape, place it in the basement, and lock it tight', they immediately have to follow this statement with '...and remember to test your backups'. When people inevitably fail to test their backups, the blame is laid on the people taking backups, not on the paradigm itself.
Consider, by comparison, git
.
Every software project gets copied from one machine to another, using git
.
As a result, they cannot ever be lost.
So we should have two separate paradigms: if you only need data integrity and don't want to use anything, it just needs checked with a hash sum, and you can lock another copy in the basement every few years. But for useful projects, for code or databases, or anything you actually work with, synchronization seems to do a much better job.