On Fri, Jan 13, 2012 at 04:20:07PM +0200, Vitali wrote:
> >> There is one more philosophical side effect of this question - speed.
> >> The closer the partition is placed to the outer cylinders, the faster
> >> the data are read from it.
> >
> > More a methaphysical question. On modern disks, the correspondence
> > between block/cyl number and physcial location is very blurred.
> >
> > B B B B -Otto
>
> Yeah, Otto, I'm aware of it, and still that makes me experiment every
> time I got an opportunity to.
> Anyway, you can use your system for years, upload and delete little
> and large files, but defragmentation ratio is very close to 0.0%.
> Either the system is not aware of this location blur or really the
> system sees that the file blocks location is indeed "convenient" for
> the system.
> In my experiments I saw large avi files being copied to the "outsider"
> /usr from a flash device 4.5mb p/s, and to the "insider" /var slower -
> depending on the size of the /usr - from 2.7mb p/s to 3.2mb p/s.
>
> I'm not insisting, I'm only telling about the results of my
> experiments... :) I know very little about physical design of HDD's
> and the vendors do not feel like sharing that information. :)
Hww scientific, conduct one experiment and distill a general rule from
it.
-Otto
