Hilmar,

I really appriciate your attempt to write an article of this subject. A few comments.

As I have mentioned before, one advantage of natural keys over surrogate keys is that natural keys make a table more readable because they have a business meaning. IMO, this is important for:

1. End - users, doing ad-hoc reports.
2. Debugging your system when something goes wrong.
3. For other developers: Readability: Not only programming code should be readable to make it easy to read for other developers. It also applies to your data !

Second, I really miss the argument that for data retrieving (whether its for reports, views or other purposes) generally more joins are neccesary and this might seriously burden performance or will face the 10 join limit in SQL commands.

Third. I don't see mentioned that for tables that not have childs and/or may have a composite key, a surrogate key is purely overhead and only takes up space (See my postcode table example).


Walter,