As mentioned in the other comment, I was actually only thinking about use cases like Athenta + csv_on_s3. and that's the really "low cost" approach I was referring to for analytical data.