[28225] in Kerberos
Re: AFS and kerberos
daemon@ATHENA.MIT.EDU (Volkmar Glauche)
Wed Aug 8 08:15:25 2007
Date: Wed, 8 Aug 2007 14:13:51 +0200 (CEST)
From: Volkmar Glauche <volkmar.glauche@uniklinik-freiburg.de>
To: Kerberos@mit.edu
In-Reply-To: <dvjua3ls1d85id09fgti7rhijnl9b0i783@4ax.com>
Message-ID: <Pine.LNX.4.64.0708081407300.21180@nz23023.ukl.uni-freiburg.de>
MIME-Version: 1.0
Reply-To: Volkmar Glauche <volkmar.glauche@uniklinik-freiburg.de>
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Errors-To: kerberos-bounces@mit.edu
Performance clearly depends on your data usage scheme... If your data is
write-once-read-many, then AFS may actually help you quite a lot. In our
environment (identical data is read approximately 5 times, but written
only once) average performance for applications is better than NFS by a
factor of 1.5-2. If your application does not read any data, but only
produces new data, then NFS is clearly faster.
Volkmar
On Tue, 31 Jul 2007, Faeandar wrote:
> AFS is great at access control and availability but it is horrid on
> performance. Read is fine if you are already cached on the client but
> we've seen a 20X performance difference between reading from AFS
> servers and NFS servers (rsync'ing data).
>
> So while AFS would solve my access control issue it would kill our
> ability to get work done.
>
> You have provided good information though, thanks.
>
> ~F
> ________________________________________________
> Kerberos mailing list Kerberos@mit.edu
> https://mailman.mit.edu/mailman/listinfo/kerberos
>
--
Volkmar Glauche
-
Department of Neurology volkmar.glauche@uniklinik-freiburg.de
Universitaetsklinikum Freiburg Phone 49(0)761-270-5331
Breisacher Str. 64 Fax 49(0)761-270-5416
79106 Freiburg http://fbi.uniklinik-freiburg.de/
________________________________________________
Kerberos mailing list Kerberos@mit.edu
https://mailman.mit.edu/mailman/listinfo/kerberos
