xfs
[Top] [All Lists]

Re: kiocluster

To: Dirk Steinberg <dws@xxxxxxxxxxxxxxxx>
Subject: Re: kiocluster
From: "Martin K. Petersen" <mkp@xxxxxxxxxxxxx>
Date: 24 Jun 2001 11:00:30 -0400
Cc: Seth Mos <knuffie@xxxxxxxxx>, linux-xfs <linux-xfs@xxxxxxxxxxx>
In-reply-to: <3B35F83B.39E43E35@dirksteinberg.de>
Organization: Linuxcare, Inc.
References: <Pine.BSI.4.10.10106232220210.18631-100000@xs3.xs4all.nl> <3B35F83B.39E43E35@dirksteinberg.de>
Sender: owner-linux-xfs@xxxxxxxxxxx
User-agent: Gnus/5.0808 (Gnus v5.8.8) XEmacs/21.4 (Copyleft)
>>>>> "Dirk" == Dirk Steinberg <dws@xxxxxxxxxxxxxxxx> writes:

Dirk> What about kio? I see it's implemented for both SCSI and EIDE.
Dirk> How about MD and LVM? 

It's implemented for MD RAID1 (by Marcelo) and I implemented it for MD
RAID0 + LVM.

We decided to pull the kiobuf support from the tree mostly due to all
the controversy about it on linux-kernel.  However, we still have the
patches kicking around so we can adapt them to whatever I/O entity
will be agreed upon for 2.5.


Dirk> Should I expect a significant performance penalty for using XFS
Dirk> on top of LVM or MD instead of running it directly on the
Dirk> physical disk partition?

Chances are your CPUs are so fast that the overhead is negligible.

-- 
Martin K. Petersen, Principal Linux Consultant, Linuxcare, Inc.
mkp@xxxxxxxxxxxxx, http://www.linuxcare.com/
SGI XFS for Linux Developer, http://oss.sgi.com/projects/xfs/

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