Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 3 Dec 2009 15:08:59 -0700
From:      Will Andrews <will@firepipe.net>
To:        Pawel Jakub Dawidek <pjd@FreeBSD.org>
Cc:        svn-src-head@freebsd.org, Alexander Motin <mav@FreeBSD.org>, src-committers@freebsd.org, svn-src-all@freebsd.org
Subject:   Re: svn commit: r200086 - head/sys/geom/mirror
Message-ID:  <20091203220856.GA73255@cephei.firepipe.net>
In-Reply-To: <20091203215532.GA2066@garage.freebsd.pl>
References:  <200912032147.nB3Llph1027086@svn.freebsd.org> <20091203215532.GA2066@garage.freebsd.pl>

next in thread | previous in thread | raw e-mail | index | archive | help

--7JfCtLOvnd9MIVvH
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Thu, Dec 03, 2009 at 10:55:33PM +0100, Pawel Jakub Dawidek wrote:
> On Thu, Dec 03, 2009 at 09:47:51PM +0000, Alexander Motin wrote:
> > Author: mav
> > Date: Thu Dec  3 21:47:51 2009
> > New Revision: 200086
> > URL: http://svn.freebsd.org/changeset/base/200086
> >=20
> > Log:
> >   Change 'load' balancing mode algorithm:
> >   - Instead of measuring last request execution time for each drive and
> >   choosing one with smallest time, use averaged number of requests, run=
ning
> >   on each drive. This information is more accurate and timely. It allow=
s to
> >   distribute load between drives in more even and predictable way.
> >   - For each drive track offset of the last submitted request. If new r=
equest
> >   offset matches previous one or close for some drive, prefer that driv=
e.
> >   It allows to significantly speedup simultaneous sequential reads.
> >  =20
> >   PR:		kern/113885
> >   Reviewed by:	sobomax
>=20
> Hmm, Will send me improved patch few days ago related to this
> functionality. Could you guys compare the patches, select the best one
> and commit it (or both if they serve different purposes)? I'm pretty
> occupied currently and it will take probably few weeks before I can join
> the discussion.

Mine changes the round-robin algorithm instead of the load one.  But mav's
patch uses a different method for distributing the load on the providers,
whereas mine is just a better-written version of the PR's patch.

--=20
wca

--7JfCtLOvnd9MIVvH
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.13 (FreeBSD)

iD8DBQFLGDb3F47idPgWcsURAsYQAJwP/0R5Vfw2OHil7iWDxkBdvb721QCfQsrx
L/BI25kd0XKQuPqE7rNgAF0=
=mPrn
-----END PGP SIGNATURE-----

--7JfCtLOvnd9MIVvH--



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20091203220856.GA73255>