disk fragmentation

disk fragmentation

[Luca Ferrari]

Hi,
I've got a simple question about disk use under windows and unix. While 
windows sometimes requires a de-fragmentation of the disk, it seems as Linux 
(and even Unix) does not. I believe this is due to a better defragmentation 
alghoritm, but I'm not sure. Is there a daemon which does this transparently 
or what?

Thanks,
Luca

-- 
Luca Ferrari,
fluca1978@virgilio.it

Re: disk fragmentation

[Chuck Campbell]

On Mon, Apr 26, 2004 at 10:18:35AM +0200, Luca Ferrari wrote:
> Hi,
> I've got a simple question about disk use under windows and unix. While 
> windows sometimes requires a de-fragmentation of the disk, it seems as Linux 
> (and even Unix) does not. I believe this is due to a better defragmentation 
> alghoritm, but I'm not sure. Is there a daemon which does this transparently 
> or what?

No,
Unix file systems, in general, are designed to have less performance issues 
from fragmentation.  They aren't as susceptible inherently to fragmentation 
problems.  As a result, there aren't any defrag tools around that I know of.

You can always build your own.  A backup to tape will sequence all the files
contiguously on tape/disk, and a subsequent restore will put them back
that way.

If you bother, do some benchmarks for a particular file i/o before and after
and I doubt you'll see much difference, unless your disk is very, very nearly
full.

-chuck

Re: disk fragmentation

[Ahsan Ali]

Yes, I believe that if the partitions have no more than 70% of their
capacity utilized there wont be any performance issues.

Please correct me if I am wrong because I havent done any benchmarks to
verify this.

Regards,

Ahsan Ali
----- Original Message ----- 
From: "Chuck Campbell" <campbell@accelinc.com>
To: "Luca Ferrari" <fluca1978@virgilio.it>
Cc: <linux-admin@vger.kernel.org>
Sent: Monday, April 26, 2004 9:08 PM
Subject: Re: disk fragmentation


> On Mon, Apr 26, 2004 at 10:18:35AM +0200, Luca Ferrari wrote:
> > Hi,
> > I've got a simple question about disk use under windows and unix. While
> > windows sometimes requires a de-fragmentation of the disk, it seems as
Linux
> > (and even Unix) does not. I believe this is due to a better
defragmentation
> > alghoritm, but I'm not sure. Is there a daemon which does this
transparently
> > or what?
>
> No,
> Unix file systems, in general, are designed to have less performance
issues
> from fragmentation.  They aren't as susceptible inherently to
fragmentation
> problems.  As a result, there aren't any defrag tools around that I know
of.
>
> You can always build your own.  A backup to tape will sequence all the
files
> contiguously on tape/disk, and a subsequent restore will put them back
> that way.
>
> If you bother, do some benchmarks for a particular file i/o before and
after
> and I doubt you'll see much difference, unless your disk is very, very
nearly
> full.
>
> -chuck
> -
> To unsubscribe from this list: send the line "unsubscribe linux-admin" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

Re: disk fragmentation

[Yu Chen]

So does this mean if the partition gets more than 70% full, I will get
performance hit?

Thanks.
Chen


On Mon, 26 Apr 2004, Ahsan Ali wrote:

> Yes, I believe that if the partitions have no more than 70% of their
> capacity utilized there wont be any performance issues.
>
> Please correct me if I am wrong because I havent done any benchmarks to
> verify this.
>
> Regards,
>
> Ahsan Ali
> ----- Original Message -----
> From: "Chuck Campbell" <campbell@accelinc.com>
> To: "Luca Ferrari" <fluca1978@virgilio.it>
> Cc: <linux-admin@vger.kernel.org>
> Sent: Monday, April 26, 2004 9:08 PM
> Subject: Re: disk fragmentation
>
>
> > On Mon, Apr 26, 2004 at 10:18:35AM +0200, Luca Ferrari wrote:
> > > Hi,
> > > I've got a simple question about disk use under windows and unix. While
> > > windows sometimes requires a de-fragmentation of the disk, it seems as
> Linux
> > > (and even Unix) does not. I believe this is due to a better
> defragmentation
> > > alghoritm, but I'm not sure. Is there a daemon which does this
> transparently
> > > or what?
> >
> > No,
> > Unix file systems, in general, are designed to have less performance
> issues
> > from fragmentation.  They aren't as susceptible inherently to
> fragmentation
> > problems.  As a result, there aren't any defrag tools around that I know
> of.
> >
> > You can always build your own.  A backup to tape will sequence all the
> files
> > contiguously on tape/disk, and a subsequent restore will put them back
> > that way.
> >
> > If you bother, do some benchmarks for a particular file i/o before and
> after
> > and I doubt you'll see much difference, unless your disk is very, very
> nearly
> > full.
> >
> > -chuck
> > -
> > To unsubscribe from this list: send the line "unsubscribe linux-admin" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> >
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-admin" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

Re: disk fragmentation

[Chuck Campbell]

> On Mon, 26 Apr 2004, Ahsan Ali wrote:
> 
> > Yes, I believe that if the partitions have no more than 70% of their
> > capacity utilized there wont be any performance issues.
> >
> > Please correct me if I am wrong because I havent done any benchmarks to
> > verify this.
> >
> > Regards,
> >
> > Ahsan Ali

On Mon, Apr 26, 2004 at 02:42:21PM -0400, Yu Chen wrote:
> So does this mean if the partition gets more than 70% full, I will get
> performance hit?

I re-ordered your question, and snipped the rest.  Top posting makes it nearly
impossible to follow, respond sanely to a thread of any length.

In line posting is chronological, in the natural reading order, top to bottom.

In answer to your question, my personal experience shows that I often get 
above 90% disk capacity without noticing any speed issues, but it is very
file size and work load dependent.  YMMV.

-chuck

Re: disk fragmentation

[Yu Chen]

> 
> I re-ordered your question, and snipped the rest.  Top posting makes it nearly
> impossible to follow, respond sanely to a thread of any length.
> 

Chunk, thankds for pointing that out, and sorry about my previous post, 
it's my misunderstanding. I thought "top posting" referring to put old 
emails on top of my reply due to my not so good English. Will follow the 
rule now. 

Chen
-- 

===========================================
Yu Chen
Howard Hughes Medical Institute
Chemistry Building, Rm 182
University of Maryland at Baltimore County
1000 Hilltop Circle
Baltimore, MD 21250

phone: 	(410)455-6347 (primary)
	(410)455-2718 (secondary)
fax: 	(410)455-1174
email: 	chen@hhmi.umbc.edu
===========================================