I'm surprised it's taken this long for a mainstream e-mail system to be based on a relational database model. Nelson E-mail Organizer (NEO) extended Outlook with this functionality a year or two ago, but I wouldn't call it mainstream, and although great software, it doesn't work well in certain environments.
Maybe as Microsoft continues development on a database-driven file system, they'll also start moving Outlook in the RDB direction. It makes so much more sense as a paradigm than the folder model.
What's the advantages of a DB driven email system?
Messages can be organized and found much more efficiently. For example, in NEO, there are preset views by Sender (in which you can group multiple e-mail addresses into a single person), messages with attachments, messages sent today, yesterday, this week, and this month, etc. There is also a full-text index, so searching for a message takes almost no time. There is no need for folders, because instant searching and multiple forms of indexing make it unnecessary.
Eh, those all seem like quanitative changes, not qualitative; all things that can be done independent of backend implementation, albeit at varying levels of efficiency.
I kind of like gmail's use of tags in lieu of folders...and the nice thing about tags is one email can have multiple tags. Come to think of it, that might be one of the nicest theoretical advantages, even if in practice the improvements in searching makes them less relevant.