evolution of the 6-bone

[Philip Blundell]

On Mon, 25 Nov 1996, Matt Crawford wrote:

> I can't quite accept your definition of Core Node, and I don't like
> the name Leaf Node as you use it.  Specifically, I don't think a core
> node should be required to provide a tunnel to everyone who asks.
> Rather, the set of all Core Nodes should be defined so that at least
> one will provide a tunnel to anyone who asks, and possibly the
> connecting site occasionally will be required to re-home when the
> Core reorganizes.
> 
> A Node which is not a Core Node may be an IPv6 provider to many other
> sites.  I think the term "Leaf" isn't appropriate in this case.
> 
> Also, you may want to require that any two Core Nodes are connected
> by an IPv6 path which includes only Core Nodes.

I agree with Matt.  I don't think it's reasonable to require any
particular Core Node to accept a tunnel from all comers.  A "leaf" node
would have to be a node which doesn't forward traffic for _any_ other
node, but I'm not at all convinced that it would be a useful concept to
come up with.  In the UK, at least, things seem to be heading towards a
loosely-organised but quite well-connected web, with many tunnels between
sites, rather than one central 'backbone' node and many 'leaf' nodes
hanging off it.

I suggest that core nodes (as defined above) be designated as such on the
map, and all other nodes just be left without any special annotation.

phil