I've just finished writing my final post in the series of 'data nirvana' posts - you can read it here tomorrow - and taken a quick look back through the other blogs I enjoy to find Rich talking about data classification being dead. I have to agree. I started writing about this last year and even ranted at someone else about not understanding it properly (which I won't dig up again).
Data classification is the real data nirvana of course, but it really can't be achieved satisfactorily. To echo Mr. Mogull for a moment, a network is a dynamic thing, it's constantly being updated with information, which can change its status from Top Secret to Private, or Public to Classified in a stroke. Tags just don't cut it. A company I spoke to at length last year propose a data classification solution. They haven't pushed it as such yet because the market isn't there. A few tyre kickers have had a go, not because they want to classify their data, but because they want to find it. That's a totally different matter. De-duplication is a very good idea, and simple, and sellable. Data classification is a great idea, but complex and completely un-sellable to anyone except me and Rich. [If you manage to invent it, please drop us a line.]
The only way you could manage to classify a system is to close it: make it read-only, or take it off-line as Rich also talks about. That kind of makes technology about as useful as your local library, though, and sends us crashing back into the 20th century just as everyone is getting used to the 21st. Something I find much more interesting is the idea of controlling information from a central hub, with policies in place around it - information sharing. It's more of a 'real world' example of how people are likely to use data security.
It reduces the need for classification as you only have to choose policies around the data you are making available outside your network. I also talked about this last year, as Microsoft released their SISA idea with about 10 other companies involved. This is clearly a good idea, but with so many technologies involved, bound for disaster. I don't know if anyone got anywhere close to deploying this, but I rather think not.
So Information Sharing is my new proxy-nirvana, or pseudo-nirvana, that is, the thing that will sell and be used, and is actually practical and possible. And guess what, I just happen to have written something about it in my post tomorrow... read on.