Feasability of VERY large ext3 file system?

Rechenberg, Andrew arechenberg at shermfin.com
Thu Dec 5 07:02:01 CST 2002


I posed this question (data=ordered being faster) on the ext3 mailing
list a few weeks ago and Stephen C. Tweedie (I believe one of the
developers, or THE developer of ext3) replied with the following:

> [ext3 is] particularly suited when you have applications performing
on-disk
> transactions --- ie. when they are using O_SYNC or fsync() to flush
> data to disk.  The data-journaling allows the data to be written
> sequentially in the log.

> However, _eventually_ it will still need to write the data elsewhere
> on disk (unless it rapidly gets deleted), so the effect is really to
> mitigate the cost of the application synchronisation, rather than
> anything else.

> So it really depends very much on what the application is doing
> whether or not this makes much of a difference.

> Cheers,
>  Stephen

You'll have to see what your application is doing WRT on-disk
transactions to see if data=journal will give you any performance
increase.

Regards,
Andy.


-----Original Message-----
From: nathan r. hruby [mailto:nathan at drama.uga.edu] 
Sent: Wednesday, December 04, 2002 5:20 PM
To: jason andrade
Cc: Ed Griffin; Steve_Boley at dell.com; linux-poweredge at dell.com
Subject: RE: Feasability of VERY large ext3 file system?


On Thu, 5 Dec 2002, jason andrade wrote:

> On Wed, 4 Dec 2002, Ed Griffin wrote:
> 
> > can the filesystem options you mentioned (rw,data=writeback) be used
for
> > these???  From what I remember when setting these systems up, I
didn't
> > change any parameters of the RAID controllers themselves, but I do
remember
> > something about a writeback option.  Should I have???
> 
> someone correct me if i'm wrong please, but i thought the above is
changing
> the way the journal works so you are increasing your risk in the event
of
> unexpected power outage (unlikely i know nowadays with dual power
supplies
> into different UPSes on different circuits..).  personally with the
large
> filesystems i am more cautious/conservative than normal because of the
> much larger data set that can be lost now (and the time to recover
from said)
> 

It is, but only a little.  Ordered mode doesn't make any promises about 
data intergrity it just does its darned best to keep your data happy.

Ironically, I found this article that states the oppisite (data=journal
is 
faster than ordered and writeback)
http://www-106.ibm.com/developerworks/opensource/library/l-fs8.html?dwzo
ne=opensource

If anyone has any comments, thoughts or experince on the above link, I'd

be interested to hear them.

-n
-- 
----------------------------------------
nathan hruby <nathan at drama.uga.edu>
computer services specialist
uga drama
http://www.drama.uga.edu/support/
----------------------------------------


_______________________________________________
Linux-PowerEdge mailing list
Linux-PowerEdge at dell.com
http://lists.us.dell.com/mailman/listinfo/linux-poweredge
Please read the FAQ at http://lists.us.dell.com/faq or search the list
archives at http://lists.us.dell.com/htdig/




More information about the Linux-PowerEdge mailing list