Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 16 Feb 2005 13:12:08 +0000
From:      David Rhodus <sdrhodus@gmail.com>
To:        delphij@delphij.net
Cc:        Scott Long <scottl@freebsd.org>
Subject:   Re: Journalling FS and Soft Updates comparision
Message-ID:  <fe77c96b05021605124959070b@mail.gmail.com>
In-Reply-To: <1108525810.676.21.camel@spirit>
References:  <20050210030119.GD29396@alzatex.com> <fe77c96b0502151935306e9aa1@mail.gmail.com> <1108525810.676.21.camel@spirit>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 16 Feb 2005 11:50:10 +0800, Xin LI <delphij@frontfree.net> wrote:
> Hi, David,
>=20
> =E5=9C=A8 2005-02-15=E4=BA=8C=E7=9A=84 22:35 -0500=EF=BC=8CDavid Rhodus=
=E5=86=99=E9=81=93=EF=BC=9A
> > > Softupdates really aren't a whole lot different from journalling.  Bo=
th
> >
> > No, journalling and soft updates are orthogonal technologies; they
> > do not solve the same problem space, although there is some minor
> > overlap.  Soft updates is not able to solve all the problems which
> > journaling can.
>=20
> Would you please give some examples?  I think a file system journalling
> is different from what we have seen in applications, e.g. databases,
> since the primary goal of having a journal is to make the file system
> recoverable to last checkpoint, is that correct, or maybe I have missed
> something?
>=20
> Thanks!
>=20
> Cheers,
> --
> Xin LI <delphij delphij net>  http://www.delphij.net/

Don't try and think of a filesystem as a database, in practice a
filesystem is not a database.

If you consider all filesystems in the context of a graph, then what
you end up with is:
1.) a graph
2.) a list of node relationships
3.) a finite state automation model for node conflict resolution


Though here is a example of why a journalled filesystem performs
betters.  With todays disk drives a DC power failure can result in an
entire track getting trashed.  Soft updates optimizes for sector
writing not track aligned writing.  Journalling can journal on the
basis of track sized extents.  This is one case were a journalled
filesystem can recover whereas soft updates could not, from a power
failure, if resolution is needed.  If this were to happen on a disk
protected by soft updates, damage may have accrued.

--=20
                                            -David
                                            Steven David Rhodus
                                            <drhodus@machdep.com>



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?fe77c96b05021605124959070b>