This sounds a lot like something I've wanted for a looong time, and that was supposedly going to be coming in Vista (at the time Longhorn), but has now been postponed: WinFS, the new database-driven file system. The capabilities of such a filesystem go far beyond your need of course, but it would be the ideal solution at least. I'm not even sure exactly what WinFS will consist of when it's released, after Vista, and I'm not exactly understanding how they intend to release it as some sort of patch since it's an entirely different file system - I suppose it'd come with an upgrade tool. But it definitely seems like a risky upgrade, probably the most major ever released for Windows between major Windows releases.
Anyway, this is also similar (as far as I can tell) to a combination of Spotlight (on OS X, or Google Desktop Search and similar apps for PC), and something like "persistent searches" or "search folders". A persistent search is basically just a keyword, file type, data range, or whatever that is visualized as a folder and constantly scans the file system for files that match its criteria. Any matching file shows up under that folder (and any other search it matches to). This kind of functionality already exists in Spotlight I think, and probably in Google Desktop, Locate, Coppernic, and other such tools. But I must admit I don't find any of those fast or smooth enough to work with to make it feasible. I have literally billions of files here (5TB of storage in my home network) and so indexing speed, while fast relatively speaking, is still too slow for my needs. The only solution is a true DB-driven file system.
The way I see such a file system work is essentially like a totally flat storage system with metadata/database fields being used to define any necessary organizational characteristics. Storing everything flat allows you to much more flexibly arrange data on your drive so that it's easy for the underlying hardware to find when needed (file system fragmention ought to decrease dramatically). Allowing unlimited metadata/db fields means you can organize your data in any way you want. The file comments feature discussed in this thread: http://www.donationc...12.msg18364#msg18364
is taken care of by the underlying file system itself. The advantages of this are tremendous and really the only disadvantage is the need for a bit faster system to handle all the DB transactions when dealing with your file system. But done intelligently the performance hit would be minimal and we have systems plenty fast enough to handle it these days.
So, when do we get our DB-driven file system MS? :-p