Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 22 Jan 1999 12:32:44 +0100
From:      "Arjan" <lists@alpha.nl>
To:        <freebsd-questions@FreeBSD.ORG>
Subject:   Re: Disk fragmentation and FreeBSD
Message-ID:  <021101be45fa$eddae9c0$98626dc2@lisboa.nedstat.nl>

next in thread | raw e-mail | index | archive | help
-----Original Message-----
From: Konrad Heuer <kheuer@gwdu60.gwdg.de>
To: Robin Huiser <R.Huiser@Winthesis.com>
Cc: 'freebsd-questions@freebsd.org' <freebsd-questions@FreeBSD.ORG>
Date: Friday, January 22, 1999 11:48 AM
Subject: Re: Disk fragmentation and FreeBSD


>
>On Fri, 22 Jan 1999, Robin Huiser wrote:
>
>> I am a using Microsoft NT Server/workstation (please read further!! :-) )
>> and I was wondering why MS NTFS suffers of extreme fragmentation while my
>> FreeBSD 2.2.6 server (which has the same amount of disk/file changes as
the
>> NT Server) has a fragmentation level of 0.6 %. (NTFS: 100 % according to
>> Diskkeeper).
>>
>> What makes the difference, and... are there even any defrag tools for
>> FreeBSD. (I don't need them, just curious!).
>
>I don't know many details about NTFS; I will limit myself to FreeBSD and
>FAT:
>
>FreeBSD uses the Berkeley Fast Filesystem (FFS). The usage of the term
>`fragmentation' concerning FFS is very different from what is expected by
>someone who is accustomed to the DOS/Windows FAT filesystem.
>
>Fragmentation in the FAT system means the arbitrary spreading of data
>blocks over the disk after a long time of operation. The result is
>limited performance. You surely know about that.
>
>The FFS doens't significantly suffer from that kind of fragmentation as
>long as there enough free space (about 10% of total space) in the file
>system since the block allocation algorithms are excellent.
>
>In the FFS, file data may be stored not only in total disk blocks (usually
>8K), but also in fragments of a block (usually 1K). To be precise, the
>last data block may be allocated imcomplete (one or more fragments). This
>is to prevent wasting of disk space by small files. For example, 8 files
>each less than 1K are stored altogether in one disk block.
>
>Thus, the rate of fragmentation of a FFS has to do with the ratio of
>allocation of fragements and blocks.
>
>Regards
>
>//
>// Konrad Heuer                                  ____            ___
_______
>// Gesellschaft für wissenschaftliche           / __/______ ___ / _ )/ __/
_ \
>//    Datenverarbeitung mbH GÖttingen          / _// __/ -_) -_) _  |\ \/
// /
>// Am Faßberg, D-37077 GÖttingen              /_/ /_/
\__/\__/____/___/____/
>// Deutschland (Germany)                      ----- The Power to
Serve -----
>//                                                http://www.freebsd.org
>// kheuer@gwdu60.gwdg.de
>//
>
>
>To Unsubscribe: send mail to majordomo@FreeBSD.org
>with "unsubscribe freebsd-questions" in the body of the message
>

In the past I've read that Linux ext2 would be good at doing this too. I'm
working with lots of small files, which have a line of approx. hundred bytes
added every once in a while. Though the story is that Linux ext2 is good at
this I am experiencing a fragmentation of +/- 20% on this particular
partition. In the near future I'm going to do some performance tests of
FreeBSD vs. Linux. Would be nice to see which of those two performs better
also regarding disk fragmentation, file access times, etc.

If anyone would be willing to give more background information, I would be
very happy to read more about the subject.


Thanks,

Arjan


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?021101be45fa$eddae9c0$98626dc2>