xfs
[Top] [All Lists]

Re: kernel spam when mounting xfs

To: linux-xfs@xxxxxxxxxxx
Subject: Re: kernel spam when mounting xfs
From: Florin Andrei <florin@xxxxxxx>
Date: 06 Sep 2001 16:39:42 -0700
In-reply-to: <Pine.LNX.4.33.0109052045580.16091-100000@xxxxxxxxxxxxxxxxxxxxxxxxx>
References: <Pine.LNX.4.33.0109052045580.16091-100000@xxxxxxxxxxxxxxxxxxxxxxxxx>
Sender: owner-linux-xfs@xxxxxxxxxxx
On 05 Sep 2001 20:53:19 +0800, Federico Sevilla III wrote:
> 
> I made a small talk for the local Linux10 celebrations. If you're
> interested you can go to
> <http://jijo.leathercollection.ph/linux10/filesystems/> to take a look at
> my slides.

Just to point out a small thing: in your presentation, you say "full
data journalling offers a severe performance impact, effectively halfing
write performance".
Well, this may or may not be true. In fact, for some workloads, Ext3 is
faster than Ext2 even with full data journalling turned on:

http://marc.theaimsgroup.com/?l=postfix-users&m=99654359818329&w=2

<quote>
- full-journalling ext3 can offer a 3x to 10x improvement over ext2,
depending upon how ext2 is used and the directory layout/task count
</quote>

(it's actually Postfix having its mail spool directory on Ext2/3, and it
seems to be true only for such usage patterns as with MTAs)

Thanks to ppetru@xxxxxxxxxx for pointing out this link.

But, anyway, your presentation is great. ;-)

-- 
Florin Andrei

"Our kernel does have source control: its name is
Linus Torvalds, CVS with a brain." - Nicholas Knight


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