@byte There's one advantage, in a technical sense: it allows for deduplicating data performantly.
Of course in practice that mostly benefits AWS and the like, because they're the ones who get the savings (and you pay full price), and there's no reason you couldn't implement a file tree structure *on top* of that...