Empirical evidence is incredibly useful for understanding your position on architecture, in deciding what the best maintenance plan for your SQL Server should be or justifying the purchase third party software for your environment.
If you haven’t seen Linchi Shea’s post series on file fragmentation and SAN, then I highly recommend perusing through the material and numbers she provides.
Hopefully we don’t have to remind anyone, but physical file fragmentation on local drives can be disastrous for SQL Server performance; for direct attached storage, it is a necessary part of any maintenance plan. Linchi’s second post in the series gives you a numerical hint when this situation goes ignored.
What is the take away from this series? Linchi talks a lot about the software vendors and the defragmentation tools as the "antagonists" in the six post plot, however, the main drive is to raise awareness so that you do not purchase a tool for the wrong situation or need and that more importantly, anything you do buy, is implemented in a more precise manner.
Take the time and read the series, it really is a great set of research posts!