--4Y142/9l9nQlBiaj
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Thu, Dec 13, 2001 at 04:22:54PM +0100, Wojciech Puchar wrote:
> really? i tried LFS for /usr and it was at least not slower (IMHO little
> faster)

I wasn't exactly thinking when I wrote that. LFS should be (and
probably actually is) faster to read /usr for the same reason it's
faster to write in the general case. You've only written /usr once,
which means that all of its blocks are written in a consecutive line
around the disk which makes your seek times (probably) be shorter
than with FFS. If you update individual files regularly in /usr,
you'll lose some speed. Probably won't see much of a hit if you
overwrite all of /user in one shot.

The only place runs into LFS reading speed issues (and they're
probably not noticeably worsed than FFS) is on file systems that
would ordinarily become fragmented (/home), since, for at least some
of the time, files larger than one disk block that have been opened,
changed, and rewritten will not be on consecutive blocks. (The
lfs_cleanerd comes through and tries to cluster those blocks
sensibly, speeding things up again somewhat, though probably not
putting files back together again completely.)

--=20
gabriel rosenkoetter
gr@eclipsed.net

--4Y142/9l9nQlBiaj
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (NetBSD)
Comment: For info see http://www.gnupg.org

iEYEARECAAYFAjwYywEACgkQ9ehacAz5CRqH1wCcC2Qm+jjPD4KPBq5X4lzUypA3
mXMAn35wXt3LGJ4r4XY3pyqoxSFf4r4e
=PoyO
-----END PGP SIGNATURE-----

--4Y142/9l9nQlBiaj--