On Fri, 23 Feb 2001, you wrote:
> neuroticimbecile wrote:
> > inode size = minimum file size
> >
> > so even if you have a 1k sized file, if your inodes are 4k, it ends up using
> > 4k of space.
>
> I see... with the way you described it, the inode size affects the system in a
> way that is similar to the cluster size of a system running DOS/Win9x...
>
> In Dos/Win9x, the smaller cluster size would contribute to more fragmented files
> in the system... would this occur on a Linux system whose inode size is set to
> 1K if large files are stored on the disk?
>
> thanks for the info...
this is something i picked up in a list once:
<snip>
That's only partially the truth. It is true that fragmentation can
not be avoided (especially in the extreme cases you describe), but
the way ext2 handles fragmentation is better than for other file
systems (well, at least some of them, i don't know about all the
fs...). The ext2 fs is designed to keep a list of all empty blocks
on the hd. When some blocks are needed to write a file, ext2 look
at his list and find *contiguous* free blocks to handle the write
(if those blocks exist, at least). The fat fs by opposition writes
the file in the first free blocks it encounters. So fragmentation
can be held at a lower rate with ext2.
Of course, as i'm not a specialist in fs, this story must only be
a small part of the truth. There is an interesting article on the
net by Roberto Di Cosmo who deals (among other things) about fs and
fragmentation. You can find the english version at :
http://www.pps.jussieu.fr/~dicosmo/Piege/PiegeEN.html
<snip>
hth,
-eric
--
.--. Enrique D. Rosel II office://+63.2.894.3592/
( () ) Q Linux Solutions, Inc.
`--\\ A Philippine Open Source Solutions Co. http://www.q-linux.com/
_
Philippine Linux Users Group. Web site and archives at http://plug.linux.org.ph
To leave: send "unsubscribe" in the body to [EMAIL PROTECTED]
