Newbie Info
John Wendel
john.wendel at metnet.navy.mil
Fri Oct 19 23:29:42 UTC 2007
Ed Greshko wrote:
> John Wendel wrote:
>> aragonx at dcsnow.com wrote:
>>>> As you can gather from the other responses. defrag isn't used often on a
>>>> Linux system. It isn't needed; the Linux kernel and friends do much
>>>> better job of allocating space on a hard disk.
>>>>
>>>> The following two sites begin the explanation. You can explore more
>>>> from there.
>>>>
>>>> http://en.wikipedia.org/wiki/Defragmentation#_note-4
>>>> http://pl.atyp.us/wordpress/?p=241
>>> I have had several threads on this very list about this subject. The
>>> threads went something like this (dramatization):
>>>
>>> Me: I want to defrag my disk
>>>
>>> Everyone else: Shut up noob. Linux doesn't need to do that winblows M$
>>> nonsense.
>>>
>>> Me: How could a filesystem never get fragmented. It's not possible.
>>>
>>> Everyone else: Go back to M$ you. We will have none of this Linux
>>> bashing.
>>>
>>> Fast forward to today, a few years later. I've got several file systems
>>> that are showing 20 - 30% non-contiguous as reported by fsck and I
>>> have no
>>> way other than a cp or dump to fix it. This just doesn't seem right.
>>>
>>> The real question is not so much if there is fragmentation or not. It
>>> is,
>>> is there a performance difference or not. Honestly, I can't say for
>>> sure.
>>> I don't have reliable tests from years ago as to how long it took to
>>> copy, write or read from a file to compare it to. I can say this, my
>>> performance is not bad enough to warrant the file system dump and
>>> restore.
>>>
>>> So, in conclusion, after years of heavy usage, my file systems still
>>> perform reasonably well. It would be nice to be able to defragment your
>>> hard drive but realistically, it isn't much needed.
>>>
>>> I hope that answers your question with real world experience.
>>>
>>> ---
>>> Will Y.
>>>
>>>
>> We haven't yet reached filesystem perfection!
>>
>> A quick google for "ext4" reveals that the file system gods think that
>> it is worth the effort of writing a new (and incompatible with ext3)
>
> Really? I thought the case was....
>
> Backward compatibility
>
> The ext4 filesystem is backward compatible with ext3, making it possible to
> mount an ext3 filesystem as ext4 (using the “ext4dev” filesystem type).
>
> Forward compatibility
>
> The ext4 filesystem is forward compatible with ext3, that is, it can be
> mounted as an ext3 partition (using “ext3” as the filesystem type when
> mounting). However, if the ext4 partition uses extents (one of the major new
> features of ext4), forward compatibility and therefore the ability to mount
> the filesystem as ext3 is lost. Extents were enabled by default in the
> 2.6.23 kernel. Previously, the “extents” option was explicitly required
> (e.g. mount /dev/sda1 /mnt/point -t ext4dev -o extents).
>
>> filesystem. The development version of the code is in the current
>> kernel.org kernel (and in F8 ???). Among its features are a new block
>> allocation scheme that keeps files contiguous and "online
>> defragmentation" (whatever that is). I assume that in a few years we'll
>> be complaining that the online defrag demon is taking too much CPU time
>> on our 64 core box.
>>
>> I amused that there is an entire industry devoted to selling software to
>> defrag winblows filesystems.
>
>
>
Excuse me, I should have read more carefully. I'll shut up now :(
Regards,
John
More information about the users
mailing list