6Bone Mail

[bound@zk3.dec.com]

My two cents on this discussion.

1. I think NHRP is a bad idea per Francis.  As Dimitry pointed out treating 
the 6bone as a LIS is not worthy of what I think its purpose is for presently.
From a networking computer science perspective though tunnels are NBMA
like, but a network 6bone tunnel to connect should not take on the semantics 
and limitations of NBMA or advantages its just a tunnel folks.

2. I think Dino's idea of static routes is a good one but agree with Craig
on 5f00/8, as 0/0 just sends everything and I don't want to debug
accidents in forwarding but only IPv6 configuration or prefix errors.
But I think we need a mechanism a.s.a.p. to discover tunnels between two
links connected by a router when one of the routers is an IPv4 node
in the way between two IPv6 islands.   I think hard conifguring this is
going to be a pain real soon.  Mike Shand had an idea and so do I so
will connect with Mike off line.  I prefer it not affecting RIPv6 but I
think it might have too.

3.  We also should not constrain again I say the 6bone and leave it open
to any who want to forward packets.  I already fear Czars who think they
might own routing in a particular region in the context of some mail.
Well just like in real life this will not fly.  We need to be prepared for a
very complicated topology where different sites will route packets
based on some relationship unforseen at this point.  No Czars.
In addition folks who build routers and forwarding code in Hosts will
want to test their implementation too.

4.  This SHOULD NOT be an IETF activity.  If we find an error in the
IPv6 portfolio of specifications then we need to take that back to the
IETF.  Lets not get bogged down in a standards game as we evolve the
6bone.

/jim