Path: utzoo!utgpu!watserv1!watmath!att!rutgers!cs.utexas.edu!samsung!uakari.primate.wisc.edu!sdd.hp.com!hplabs!hpl-opus!hpspdra!burdick
From: burdick@hpspdra.HP.COM (Matt Burdick)
Newsgroups: comp.sys.apollo
Subject: Re: Re: reason to buy Apollos, how about older used ones?
Message-ID: <13730002@hpspdra.HP.COM>
Date: 16 Jul 90 15:08:28 GMT
References: <9007070357.AA00259@zeus.WRI.com>
Organization: HP Stanford Park - Palo Alto, CA
Lines: 30

> Also like the Apollo setup, AFS will let you access directories such
> as /<hostname>/<path> (ie: /eddie.mit.edu/usr/tmp).

I just received a reply to my query in comp.soft-sys.andrew (a
newsgroup for Andrew-related topics) about AFS.  Since this person
works at Transarc and is more knowledgable about AFS, this is what
they said:
---------------------------------

Essentially, naming doesn't work that way (/dopey/usr/tmp ->
``/usr/tmp'' on machine dopey).  AFS files are in a totally separate
name space, not connected in any UFS sense to any machine's native
file system.  AFS directories aren't UFS directories.

Usually the machines that store files aren't any random workstation,
but rather dedicated, secured machines.

Now, to AFS 3.0: /afs/transarc.com/usr/cfe is my AFS home dir.
``/afs'' is the mount point off which the global name space hangs.
``transarc.com'' is the name of my cell, which represents a lot of
servers running on a lot of machines in that cell.  ``usr/cfe'' is the
path name within my cell--that is, using the root directory in the
cell's root volume to get at the directory ``usr'', and using that to
get to the directory ``cfe''.

--------------------------------------

-- 
Matt Burdick                |   Hewlett-Packard
burdick@hpspd.spd.hp.com    |   Intelligent Networks Operation