xfs
[Top] [All Lists]

Re: Fragmentation (was: XFS NFS server Oops)

To: Rupa Schomaker <rupa-list@xxxxxxxx>
Subject: Re: Fragmentation (was: XFS NFS server Oops)
From: Andrew Morton <akpm@xxxxxxxxxx>
Date: Tue, 19 Mar 2002 22:48:06 -0800
Cc: linux-xfs@xxxxxxxxxxx
References: <1016578034.28166.25.camel@xxxxxxxxxxxxxxxxxxxx> <Pine.LNX.4.44.0203200925430.8497-100000@xxxxxxxxxxxxxxxxxxxxxxxxx> <m34rjbhl3e.fsf@xxxxxxxxxxxxxxx>
Sender: owner-linux-xfs@xxxxxxxxxxx
Rupa Schomaker wrote:
> 
> Federico Sevilla III <jijo@xxxxxxxxxxxxxxxxxxxx> writes:
> 
> > I got fragmentation factors from a high of 50.92% (this is a filesystem
> > where I store huge tar.gz backups and not much else) to a low of 0.43% for
> > my / filesystem.
> 
> I've got a filesystem with 28 .iso images, the frags -f command shows:
> 
> # xfs_db -r /dev/shaktivg/xxx
> xfs_db: frag -f
> actual 355, ideal 29, fragmentation factor 91.83%
> 

I assume that's saying that ideally, those 28 files should take
up just 28 contiguous chunks of disk, but in fact they're taking
355.  If those ISOs are all 600 megs then your average contiguous
chunk size is 47 megs, which is darn good.

Quantifying fragmentation is hard.  I think the only valid
way really of measuring it is to go for the bottom line:
grab the stopwatch and time how long it takes to read the files.
Then copy them to a newly-initialised and empty filesystem, see
how long it takes to read the files under these ideal circumstances.
Then compare the two times.


<Prev in Thread] Current Thread [Next in Thread>