Hierarchal data, and relational DBs do not mix well. There are hackish solutions involving adjacency lists and several variations, but they are all inefficient and clumsy at best, and downright friggin' awkward at worst.

See MySQL:Managing hierachal data for the skinny.

It gets easier with Stored Procedures, but it can still get slow and messy. Sometimes, it is worth at least asking the question: is there a better way to persist my data than a RDBMS, and hierarchal data is one such time.

The primary questions are:

• Do you need concurrancy?
• Do you need performance?
• What volumes are you dealing with?

A realistic take on the answers to those questions can go a long way to suggesting other possible alternatives.