Excellent article, Jeff! I look forward to Thursday's installment.

I've played around with mixing these two representations of hierarchies before, but had never gotten past the performance issues, so it's great to have a new approach to try. It seems that I keep inheriting db implementations that contain at least one hierarchy in them, so having more choices for dealing with them is a very good thing!

For those that are willing to give up a little in terms of understandability (which can be dealt with via abstractions, anyway), and who are working with fairly constrained hierarchy sizes, take some time to investigate the matrix path encoding technique that Tropashko describes. You can think of it as a "best of both worlds" approach to the problem that doesn't have the heavy recalculation requirements. I was also able to obtain favorable performance results for many of the descendant- and ancestor-type queries. If nothing else, it will expand your db math skillset:-D

And finally, for those with some disk space to burn, you might consider materializing the transitive closure of the hierarchy. This can really net you some performance if you have a large hierarchy that is used primarily for reads (the materialization is a rather expensive process).

-TroyK