Re: Tunnel-to-NAT scenario

[Rémi Després <remi.despres@free.fr>]

Jari Arkko  - Le 6/17/08 4:15 PM :

> ... On initial look, your mechanism seems more of a general nature 
> than the specific case sent to SOFTWIRE.
IMHO it has more similarities with the work of SOFTWIRE than you have 
found:
- It has tunnels the endpoint of which can be in hosts or router CPEs.
- For the v4-v6 coexistence ad hoc functions are needed in both router 
CPEs and in ISP infrastructures.

The main difference with Tunnel-to-NAT, for the same 464 paths, is that 
NAT44 is in CPEs rather than in the ISP infrastructure.
- The disadvantage is more work left on specification (but IMO not so 
complex)
- The main advantage is a better scalability. (And the next advantage is 
E2E native IPv4 in client hosts that support APBP).

Anyway, it seems too early to decide where APBP would fit.
Discussing it in Dublin would IMO make sense.


Rémi