Shudder,
I feel that Norton's defragger is a triumph of obsession over reason. As you appear to be suspecting, it's for anal retentives who want the data arranged the way they think they'd like to see it arranged if they were disk drives. Diskeeper (commercial versions), O&O and Perfect Disk work better for the system's sake. (And I'm certain there may be others with which I'm not familiar.) The Norton defraggers, when set to do online MFT defragging, are also flying in the face of reason by defragging something online that should only be defragged at boot time. And Microsoft says so right in the MSKB.
shathal,
<< Nothing (or rather, very little) kills/stresses drives as much as defragging. As such, they shouldn't be defragged too often. Also, realistically speaking - drives only get badly fragmented over time. If you defrag the drives once every 3-6 months you will be fine really. >>
I've seen this assertion made many times, but I strongly disagree. And I have a great deal of experience with data on MTBF of drives in use on machines with and without the benefits of routine defragging. Defragging doesn't kill drives, nor does it necessarily stress drives any more than ordinary use does. Regarding electro-mechanical stresses induced by defragging vs. normal use patterns, there is a break-even point, probably somewhere between the home user and a small database server's usual operations in most cases, where the drive would actually be stressed more if you didn't defrag routinely. In other words, the heads have to jump about more to retrieve / write data because files are not written contiguously, so the wear and tear you save by not defragging is more than compensated for by increased wear and tear during the system's regular operations. The more heavily the system is used, the more important it is to guard agains file system fragmentation. On systems which do a lot of thrashing, like busy SQL servers, not defragging is the death of system performance -- and definitely results in far more wear and tear on the system's drives. Benchmarks prove the point about performance, and the time between failure statistics prove the point about wear and tear.
Regards,
Jim