RE: draft-huitema-v6ops-unmaneval-00.txt

["Christian Huitema" <huitema@windows.microsoft.com>]

Itojun,
 
When it comes to traversing NAT, the issue is how to "come in", not how to "go out". AFAIK, teredo is the only solution that allows "self deploy" because it let UDP packets "come in". All the other solutions are variations on fixed tunnels, and thus require some form of contract between a user and a tunnel endpoint. You can clearly do PPP over TCP, but you can only do that if you initiate the connection from inside the NAT towards a fixed tunnel end-point; in that case, I believe that a UDP base tunnel will provide a lesser overhead, and will avoid the "head of line blocking" issues found in TCP. RFC 2893 tunneling only works with some NAT, and as such cannot be a generic solution.
 
-- Christian Huitema

________________________________

From: Jun-ichiro itojun Hagino [mailto:itojun@iijlab.net]
Sent: Sat 3/15/2003 9:13 AM
To: v6ops@ops.ietf.org
Subject: draft-huitema-v6ops-unmaneval-00.txt



>2.1    Evaluation of connectivity mechanisms
>  
>   In case A, IPv6 capable hosts seek IPv6 connectivity in order to
>   participate with applications in the global IPv6 Internet. The
>   connectivity requirement can be met in two ways: Teredo [TEREDO], or
>   UDP tunnels.

        there are many other ways to solve case A, including:
        - IPv6 PPP over TCP (can penetrate any NAT as it is TCP)
        - RFC2893 tunnelling with certain NAT/decapsulator configuration
        if you wish to limit solutions to teredo and UDP tunnels, you would
        need to describe why.

>6       Provisional recommendations
>   -    To meet case A requirements, we need to develop and standardize
>        the Teredo or similar technology.

        therefore, it is too early to conclude like this.

itojun