If you’ve ever designed a database from scratch, or worked with a database migrations, then you know how important it is to get the data schema right the first time. If you get them wrong, then when you (inevitably) have to fix it, you must do a major overhaul of your code just to fit the updated schema. In Rails, this means re-implementing potentially dozens of objects. In Haskell, it’s not as bad because the compiler handles most error checking (with Persistent, anyhow), but it’s still not a trivial exercise.
So how do you design a database with Persistent? You begin with database normalization (DN) in mind.