G6, new topology & routing policy

[Alain Durand]

Hi.

I have change the topology of my IPv6 testbed and or routing  policy.
I hope this might help the discussion on the restructuration of the 6-bone.

We are now using two routers as tunnel entry points
 One is our regular sun (6bone-gw.ipv6.imag.fr/129.88.26.1)
with netbsd/inria code.
The other one is a digital router (6bone-core.ipv6.imag.fr/129.88.26.2).

The goal is to somehow split core routing from leaf access. In the migration
process, 6bone-gw will be use for regular, non-RIPng tunnels and 6bone-core
will be used for RIPng ones. 6bone-core will advertize routes for sites I have
direct non-RIPng tunnels to via 6bone-gw. I hope this 2 routers scheme will
help me to select the routes I want to advertize with RIPng.

Just to fuel the discussion, here are some details about G6 tunnels.

Non rip-ng tunnels readvertized through RIPng via 6bone-core:
WIDE/jp, UNI-C/dk, UL/pt, UNH/us, JOGUNET/de, SICS/se, COSY/at , ERA/se,
NRL/us, SUMITOM/jp and CSELT/it

RIPng tunnels:
BAY/us, DIGITAL-ETC/fr, DIGITAL-CA/us & DIGITAL-BE/be

Some non RIPng tunnels will be replaced shortly by core RIPng ones.

If we reorganize the 6-bone in a geographical approach, I realize
I should not annouce some sites and maybe I should dig some new tunnels
with other european sites.

I would like to get some comments: is a geographical approach a good thing?
will it (or should it) map the underlaying IPv4 networks?

I think this is an important point in this discussion.

	- Alain.