Once and a while I'm asked for independent studies on the performance effects of disk fragmentation. There are some excellent studies out there. Diskeeper Corporation recently sponsored a paper done by Windows IT Pro. Their writer, Joe Kinsella, actually developed his own test tools in order to do a very thorough fragmentation performance study:
http://www.windowsitpro.com/whitepapers/index.cfm?fuseaction=showwp&wpid=4D2B047D-E1F8-4F20-A74DB68E7521E730&code= Besides the recent study we sponsored, there are a bunch of other independent studies in existence. A quick Google search yields a Harvard University study on UNIX:
http://www.eecs.harvard.edu/~keith/research/tr94.html I was unable to find a university study on Windows. I imagine Universities are probably less interested in studying the effects of fragmentation on FAT and NTFS, because it's generally considered common knowledge that file fragmentation hurts their performance. Once in a while a media source does in-depth fragmentation testing as part of a review, such as this PC Mag article:
http://www.pcmag.com/article2/0,1895,1591466,00.asp The non-profit organization National Software Test Lab (NSTL) has also done performance tests on fragmentation:
http://www.softwareshelf.com/HTML/products/prod_materials/5/NSTLXP_mddvdk.pdf - Paul