From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mike Waychison Subject: Re: Performance of autofs/amd-managed directory Date: Wed, 26 May 2004 08:47:09 -0400 Sender: autofs-bounces@linux.kernel.org Message-ID: <40B491CD.1070707@sun.com> References: <20040526030234.47115.qmail@web20025.mail.yahoo.com> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: In-reply-to: <20040526030234.47115.qmail@web20025.mail.yahoo.com> List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: autofs-bounces@linux.kernel.org Content-Type: text/plain; charset="us-ascii" To: Kenny Simpson Cc: autofs@linux.kernel.org -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Kenny Simpson wrote: > What is the performance penalty for write access to files in a managed > directory vs. a fstab entry? > I have a program that is very sensitive to write latency, and am being > pressured to use amd or automount to access a NAS instead of my trusty fstab > entry. > > My app uses a transactional log file that resides on a NAS. I'd prefer to keep > a simple fstab entry and use O_DIRECT to access the file, but if there is no > penalty, I would switch. Does O_DIRECT have meaning for automount or > amd-managed files? (isn't amd user-space, so it would need to copy data > anyway?) > As Ian already pointed out, automount simply mounts filesystems in place when requested and does not affect any file access fast-paths. Similarly, amd also simply mounts the NFS filesystems, albeit in a different location and uses symlinks to redirect the user. This has the side-effect that a chdir('..') will drop you in a different directory. As long as you have O_DIRECT support in your NFS client, neither solution will have any real adverse effects on your transactional log. - -- Mike Waychison Sun Microsystems, Inc. 1 (650) 352-5299 voice 1 (416) 202-8336 voice mailto: Michael.Waychison@Sun.COM http://www.sun.com ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ NOTICE: The opinions expressed in this email are held by me, and may not represent the views of Sun Microsystems, Inc. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iD8DBQFAtJHNdQs4kOxk3/MRAuBmAJ9Ax8CPblvJOiil5CNvFJYerkyMSACeL3wI ckLOQO+1oIecf0q2Ujoqwe8= =Z5rB -----END PGP SIGNATURE-----