Re: draft-durand-dual-stack-lite-00 general comments

[Iljitsch van Beijnum <iljitsch@muada.com>]

On 27 jul 2008, at 22:19, Alain Durand wrote:

> > By the way, my Ubuntu PC refuses to accept 0.0.0.1 as a valid  
> > address. I
> > reckon the same operating system is found in certain CPEs.

> This address is mainly used in the case a device -that-is-not-behind- 
> a-CPE-
> wants to initiate the tunnel. In this mode of operation, dual-stack  
> lite is
> a very simple alternative to NAT-PT.

> Now, I suggested 0.0.0.1, but this may not be the best choice. I will
> probably end up asking IANA to reserve an address for this.

The right choice would be something in the RFC 1918 space so  
applications that inspect their source address know they're behind NAT  
and change their behavior accordingly.

In the MNAT-PT draft from seven months ago there's text about creating  
a source address in the 172.31/16 range where the lower 16 bits are  
chosen such that they compensate for the checksum difference between  
IPv4 and IPv6 so there's no need to recalculate the checksums.