Date: Fri, 23 Mar 2001 20:08:43 -0800
From: Steve Deering <[EMAIL PROTECTED]>
Message-ID: <v04220802b6e1c6d1bdbb@[10.83.97.214]>
| I have a hunch that you, and probably many others, will still find this
| description less than satisfying, preferring a simple, concrete rule for
| defining a site, but perhaps you can at least get a glimmer of the
| general notion that has been mostly buried in my mind and inadequately
| documented so far.
I find it less than satisfying, but not for that reason, in fact, for
exactly the opposite reason, that definition is far too precise.
A site should be whatever I want it to be. About the only requirement
should be that it is internally connected (somehow, including using tunnels).
No more than that is needed - it is just a collection of nodes that the
administrator defines are a site, and then configures the routers at the
borders of the collection of nodes to mark them as site boundary routers.
I can't imagine why any relationship with buildings, geography, campuses,
or anything else like that should be necessary - they add nothing.
Just as the routing architecture, etc, don't impose any requirements on
just what the thing is that gets a /48 routing prefix, beyond that it needs
to be a connected network, we also don't need any specific definition of
what is a site - all that can possibly do is to constrain thinking into
the future.
kre
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
