Apparently, this has been a bit of an issue. I did some poking around,
and, firstly, it appears that good ol' ext(1) filesystem was *very*
prone to fragmentation. Then, as this link shows, ext2 came along, and
was better, and then got tweaked, to boot:
"... Note that it is not so bad as it seems : the ext 2 fs causes
*much* less fragmentation than ext fs so there is very little need for a
defragmenter.
Moreover, Stephen Tweedie (the author of defrag and edefrag) has
rewritten the ext 2 fs allocation routines and they should produce even
less fragmentation than before in the next version. Now, thanks to
Stephen's good work, the ext 2 fs tries (and succeeds in most cases) to
allocate contigous blocks for the same file. These new allocation
routines are already included in Linus' latest Alpha kernel."
Being as this snippet was from '93 (see link, below), I guess that
things got cleaned up relatively early on. That still, of course, isn't
a benchmark, but it does show that it was a real concern that was
addressed head-on. (By Stephen Tweedie, no less, a developer for whom I
have a tremendous amount of respect -- and the initial instigator for
ext-3. Hadn't realized he'd been around the kernel that long, though.
Wow.)
So, is it a benchmark comparison? No. Does it give me warm-fuzzies?
Yup.
http://www.kclug.org/old_archives/linux-activists/1993/apr/0/0203.shtml
$.02
-Ken
On Thu, 2002-01-10 at 21:56, Jerry Feldman wrote:
> Good question Benjamin,
> >From my experience with ext2 on my systems, I have never seen significant
> fragmentation on ext2 when if was less than 80% full, but this is not
> really empiracle evidence. However, I was very disappointed with NTFS. I
> have found on my Windows 2000 system at work the need to defragment almost
> immediately after an install.
> I think that part of the difference is that Linux buffers all of its I/O
> where NT systems do not (AFAIK).
> Benjamin Scott wrote:
>
> > Does anyone know of any unbiased and current analysis of this issue? The
> > information in that HOWTO is rather biased, and contains the out-of-date
> > assumption that multi-hundred-gigabyte filesystems are unlikely. It
> > basically answers the question of "Does ext2 suffer from fragmentation
> > issues?" with "No, because MS-DOS sucks." While I agree with the author's
> > sentiments, they do not support the conclusions. At the same time, NTFS
> > supposedly has a much better design, and analysis of *it* has shown that
> > fragmentation *is* still an issue for some applications. Has anyone
> > actually tested ext2 for this, or are we just happy living in dreamland?
> >
> > --
> > Ben Scott <[EMAIL PROTECTED]>
> > | The opinions expressed in this message are those of the author and do not |
> > | necessarily represent the views or policy of any other person, entity or |
> > | organization. All information is provided without warranty of any kind. |
> >
> >
> > *****************************************************************
> > To unsubscribe from this list, send mail to [EMAIL PROTECTED]
> > with the text 'unsubscribe gnhlug' in the message body.
> > *****************************************************************
> >
>
> --
> Jerry Feldman <[EMAIL PROTECTED]>
> Boston Linux and Unix user group
> http://www.blu.org PGP key id:C5061EA9
> PGP Key fingerprint:053C 73EC 3AC1 5C44 3E14 9245 FB00 3ED5 C506 1EA9
>
>
>
> *****************************************************************
> To unsubscribe from this list, send mail to [EMAIL PROTECTED]
> with the text 'unsubscribe gnhlug' in the message body.
> *****************************************************************
>
*****************************************************************
To unsubscribe from this list, send mail to [EMAIL PROTECTED]
with the text 'unsubscribe gnhlug' in the message body.
*****************************************************************
