> >and apart from what Ann recommends, with fast HDs of today and
> NTFS, defrag
> >is nowhere near as effective as it was in the early days with
> slow HDs and
> >FAT systems.
> >It's effect is so marginal it's not really worth considering.
>
> Are you kidding? A fraggy fast disk on NTFS is accessed faster than a
> fraggy slow disk on FAT32 but it's still slowed by being fraggy.

I've done this many times sine NT4 server (but not for quite a while with
any seriousness) and in my experience, post defrag even on large dbs shows
little or no performance enhancement. Have you measured it? On ms selects
are you talking about a ms improvement?
Maybe on a system which is hammered with natural scans a lot you may see
improvement but I get far better performance enhancement from doing a little
profiling and adding an index or two where needed.
Alan


The

> impact on files that have a low level of access might be
> insignificant but,
> with a database file, disk fragmentation adds progressively more
> levels of
> indirection over and above the internal indirections in effect
> *within* the
> database file(s).
>
> One thing I have observed with fraggy NTFS disks and the shipped defrag
> utility is that you have to defrag it twice to get it done properly. At
> least, that's the case on my Win2K systems, where the RAM isn't enormous.
>
> ./heLen