This article makes no sense whatsoever. Redis is an in memory database, that's what it does. Obviously what this guy wants is something entirely different. This is like complaining that your apples lack the needed orange colour and make don't even produce orange juice!
Redis by itself is an in-memory database. The author is writing about Redis-vm, which supports paging to disk for datasets larger than ram. Basically all he's doing is recommending some alternatives to the current approach, other than writing your own btree implementation.
Redis is not just an in-memory database though. It's many things, of which persistence database (if you set it) is one. Although I disagree with the article, what this guy wants is what Redis seems to be heading towards, but as far as I know has not claimed publicly to be. He seems to be worrying too much over nothing.
