I've used this technique with success on various large databases. It has helped with the performance on lookups and the physical size of tables. However, on the downside you do increase the complexity and the time taken to write queries. I would not suggest using this method across the entire database, but rather for splitting tables which may have a lot of duplicated varchar or char fields mixed in with int, float etc. I refer to this method as partitioning tables rather than using a "store key" but from what I read in the article the concept is the same.

Overall I think if you have a lot of repeated data due to the lack of orginal thoughful database design (and it happens) this is a method that may be worth trying. Of course test everything before implementing it into a production system.

Cheers,

Angela