Re: [PATCH 01/11 RESEND] libe2p: Add new function get_fragment_score()

[Greg Freemyer <greg.freemyer@gmail.com>]

On Tue, Jun 28, 2011 at 6:21 AM, Kazuya Mio <k-mio@sx.jp.nec.com> wrote:
> 2011/06/26 11:16, Greg Freemyer wrote:
>>
>> I'm talking about sparse files.
>>
>> Where you might have:
>>
>> block_group1 - hole - block_group2 - hole - block_group3.
>>
>> Block_group2 has a head and a tail extent.  In my mind, from a
>> performance perspective, they are symmetric.  Meaning that having a
>> small extent at the beginning is no better and no worse than having a
>> small extent at the end.
>
> Thanks for your kind description. Your point is right. I'll think a little
> more
> about the fragmentation score.
>
> Regards,
> Kazuya Mio

Kazuya,

While you're thinking about the issue:

As I hope I've said before, for sparse file I think e4defrag should
score and defrag one block_group at a time.  Thus if a VM backing
storage file has 100 block_groups (as I'm using the term), then it
should score each of the 100 separately and if needed defrag them one
at a time.

I can see no benefit from treating a large sparse file as monolithic
for the decision process.

fyi: Is there an agreed on term for what I'm calling a block_group.  I
believe e4defrag uses the term "extent group" in the comments, but
sparse files exist in non-extent based filesystems, so it's not a very
portable name.

Greg
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html