Outstanding article...clear in meaning, no clutter or distractions, straight to the point.

It is also enlightening to see some of the beginner mistakes I made so many years ago when I tried to use the relational database as storage for a perfect object model and used only a few generic tables holding all the data. Your article correctly describes the problems it had when soon after it wend into production...shockingly quickly I might add, even on brutal hardware. That everything was developed on a memory constraint laptop did not help either ofcourse, always make sure to also test on similar hardware it will eventually be running on. A memory constraint laptop running 5 applications next to SQL server will not correctly separate bad code/models from good code/models. It is like trying to find out which athlete can run faster and should be send to the olympic games, with the constraint that they all have to tug 800kg of metal behind them to find it out...uphill.

As for the bitfield problem (and similar ones), filtered indexes (SQL 2008) can be a big help here.