- Subject: Re: [Lsf-pc] [LSF/MM TOPIC] a few storage topics
- From: Chris Mason <chris.mason@xxxxxxxxxx>
- Date: Tue, 24 Jan 2012 10:15:04 -0500
- Cc: Andrea Arcangeli <aarcange@xxxxxxxxxx>, Jan Kara <jack@xxxxxxx>, Mike Snitzer <snitzer@xxxxxxxxxx>, linux-scsi@xxxxxxxxxxxxxxx, dm-devel@xxxxxxxxxx, Boaz Harrosh <bharrosh@xxxxxxxxxxx>, linux-fsdevel@xxxxxxxxxxxxxxx, lsf-pc@xxxxxxxxxxxxxxxxxxxxxxxxxx
- In-reply-to: <firstname.lastname@example.org>
- Mail-followup-to: Chris Mason <chris.mason@xxxxxxxxxx>, Jeff Moyer <jmoyer@xxxxxxxxxx>, Andrea Arcangeli <aarcange@xxxxxxxxxx>, Jan Kara <jack@xxxxxxx>, Boaz Harrosh <bharrosh@xxxxxxxxxxx>, Mike Snitzer <snitzer@xxxxxxxxxx>, linux-scsi@xxxxxxxxxxxxxxx, neilb@xxxxxxx, dm-devel@xxxxxxxxxx, linux-fsdevel@xxxxxxxxxxxxxxx, lsf-pc@xxxxxxxxxxxxxxxxxxxxxxxxxx, "Darrick J. Wong" <djwong@xxxxxxxxxx>
- References: <20120117200609.GA7933@redhat.com> <20120117213648.GA9457@quack.suse.cz> <20120118225808.GA3074@tux1.beaverton.ibm.com> <20120118232200.GA22019@quack.suse.cz> <4F1758D4.email@example.com> <20120119094637.GA23442@quack.suse.cz> <4F1BFF5F.firstname.lastname@example.org> <20120123161857.GC28526@quack.suse.cz> <20120123175353.GD30782@redhat.com> <email@example.com>
- Reply-to: device-mapper development <dm-devel@xxxxxxxxxx>
- User-agent: Mutt/1.5.21 (2010-09-15)
On Mon, Jan 23, 2012 at 01:28:08PM -0500, Jeff Moyer wrote:
> Andrea Arcangeli <aarcange@xxxxxxxxxx> writes:
> > On Mon, Jan 23, 2012 at 05:18:57PM +0100, Jan Kara wrote:
> >> requst granularity. Sure, big requests will take longer to complete but
> >> maximum request size is relatively low (512k by default) so writing maximum
> >> sized request isn't that much slower than writing 4k. So it works OK in
> >> practice.
> > Totally unrelated to the writeback, but the merged big 512k requests
> > actually adds up some measurable I/O scheduler latencies and they in
> > turn slightly diminish the fairness that cfq could provide with
> > smaller max request size. Probably even more measurable with SSDs (but
> > then SSDs are even faster).
> Are you speaking from experience? If so, what workloads were negatively
> affected by merging, and how did you measure that?
This patch is another example, although for a slight different reason.
I really have no idea yet what the right answer is in a generic sense,
but you don't need a 512K request to see higher latencies from merging.
dm-devel mailing list