[OpenAFS] Re: OpenAFS-info digest, Vol 1 #406 - 2 msgs

Ted Anderson ota@transarc.com
Mon, 10 Dec 2001 10:41:45 -0500 (EST)


On Sun, 09 Dec 2001 14:06:57 -0500 Steve Burling <srb@umich.edu> wrote:
> The approach that several of us here at the University of Michigan
> have taken for this is to mount my volume (user.srb) as /afs, and then
> to create mount points in my volume for the root cell of those cells
> I'm interested in.
...
> This has the additional nice side effect that paths like
> 
>     /afs/umich.edu/<whatever>
> 
> still work properly.

This is a very cool idea!  It gives the user/client (or returns to him)
full control of his file system namespace.  I view this as one of the
philosophical weaknesses of the global naming architectures used by AFS
and DFS.  I've thought about taking a SPKI approach (where each user has
his own namespace of public keys) but didn't realized that AFS and DFS
could already support this architecture.

Excellent,
Ted Anderson