[8386] in Info-AFS_Redistribution
Re: File copying
daemon@ATHENA.MIT.EDU (Paul Blackburn)
Wed Jan 3 07:13:28 2001
Message-Id: <3A53119D.A1DED231@hursley.ibm.com>
Date: Wed, 03 Jan 2001 11:48:45 +0000
From: Paul Blackburn <mpb@hursley.ibm.com>
Mime-Version: 1.0
To: info-afs@transarc.com
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Peter,
I think you will find that if the copy is run
on the fileserver itself then the loopback
address is used (eg data does not flow
"across the wire"). This is a special
case because I would normally expect
only system administrators to have
access to the fileserver.
Have you thought of using a symbolic link
instead of copying the file?
I believe AFS was not designed to optimise 500 MB file copying
from a client machine but it does provide a scalable secure
distributed filesystem.
--
cheers
paul http://acm.org/~mpb
Peter Scott wrote:
> Something tangential to the backup product discussion that I would like to
> toss out there - if any enterprising developers would like to come up with
> a native file copy command that acts purely server side, it would be so
> great. The frustration in copying a 500MB file from one AFS location to
> another in the same cell (or at least on the same server), knowing that all
> the bytes are travelling up the wire to the poor old workstation whose only
> job is to send them back again, is palpable.
> --
> Peter Scott
> Peter.J.Scott@jpl.nasa.gov
