News.com: "To achieve the long-elusive goal of easily finding information hidden in computer files, Microsoft is returning to a decade-old idea."
First they tried to make the file system a database. That didn't work, so now they'll try to make the database a file system. Forgive me while I choke down my scepticism.
It's not that I believe it can't be done, it's just that I'm not sure it should be done. Data warehousing projects in general seem more likely to end up as runaways than most other large systems, and this plan sounds like the mother of all warehousing projects. Define a data model that can handle, uhhh, anything. Now make it super-efficient, then move all your data into it. Good luck!
The irony is that other parts of Microsoft seem to 'get it', and appear to be headed toward federated systems built from loosely connected autonomous nodes that communicate via message passing, making no assumptions about the underlying data representation at any given node: sort of like deterministic finite automata at Internet scale. This makes much more sense to me than trying to stuff everything into a giant database.
Hopefully this is an example of botched reporting in the trade press, and not a direction shift.
Time will tell, and I'll be watching anxiously and hoping for an invite to the first SDR on this stuff. [Peter Drayton's Radio Weblog]
8:31:14 PM
|