[dcc2] NAT traversal for direct connections

[Phoenix Fyrestar]

I belive we should go with STUN/UPNP.  Although I am admittedly bias, I have a 
hard time thinking much long term good can come from standardizing on 
something from microsoft as opposed to a free and open protocol.

On Friday 21 May 2004 01:03 am, Dan Smith wrote:
> Hey everyone,
>
> I have been updating the negotiation draft to address our last months
> discussion.  I think we still have more to talk about with the NAT
> traversal.
>
> The current NAT token should be renamed to PASV or something similar, to
> indicate a passive transfer would be preferred, and be added as an option
> in a "new" NAT token.  The NAT token will then contain a list of supported
> NAT traversal protocols.  Which traversal techniques do you think we should
> adopt/standardize around?
>
> The two that I am leaning toward are STUN+UPNP, and microsofts Teredo
> technique for IPv6 encapsulation over IPv4/UDP.
>
> There are advantages and disadvantages to each.  With teredo, a
> intermediate teredo server (an ipv4 and ipv6 connected computer) is
> needed.  STUN does not allow incoming tcp connections or work with
> symmetric NATs, so connections would have to use a tftp type protocol over
> udp.
>
> There is "no good way" to solve the problem of NAT traversal beside global
> addressing (which DCC2's ipv6 support addresses), but we can still solve of
> most of the NAT traversal cases.  What are your opinions/comments or other
> protocol suggestions?
>
> Quicklinks:
> http://www.microsoft.com/technet/prodtechnol/winxppro/maintain/teredo.mspx
>
> http://www.ietf.org/rfc/rfc3489.txt
>
> Cheers!
> Dan
>
>
> ---------------------------
> Dan Smith
> +1 608-213-2867
> Algenta Technologies, LLC
>
> _______________________________________________
> dcc2 mailing list
> dcc2 at dcc2.org
> http://six.pairlist.net/mailman/listinfo/dcc2