Re: [Gnu-arch-users] Some issues

[Andrew Suffield]

On Wed, Jun 09, 2004 at 07:38:09PM +0200, Florian Weimer wrote:
> * Andrew Suffield:
> 
> >> But it's so much cheaper to read the data from a single, larger file,
> >> than from lots of different ones.
> >
> > This is essentially wrong on all the interesting filesystems. It's
> > equivalent on most of them, and a little slower to use one large file
> > on NTFS and HFS+.
> 
> Huh?  What about seek overhead?

What seek overhead? There is no strong relationship between "being in
the same file" and "being in the same region of the disk", and no
strong repulsion between "being in different files" and "being in the
same region of the disk". These things are mostly related to creation
pattern, not file membership.

Interesting filesystems do not seek the head to park and back every
time they read from a different file. They actually do the opposite,
and reorder disk accesses to minimise seeks.

HFS+ goes further, and shuffles small files around so that the ones
you use most often are stored near each other; this doesn't work if
you cram them all into one huge file. (NTFS is just inefficient at
allocating large files, and tends to scatter them, while small files
are packed together more densely; yes, its allocation algorithm
sucks).

-- 
  .''`.  ** Debian GNU/Linux ** | Andrew Suffield
 : :' :  http://www.debian.org/ |
 `. `'                          |
   `-             -><-          |

signature.asc
Description: Digital signature