Re: [BEHAVE] Comments on the NAT66 draft

[Rémi Després <[e-mail]>]

Gert Doering   (1-12/1-31/200x) 11/10/08 4:19 PM:

Hi,

On Mon, Nov 10, 2008 at 08:16:04AM -0700, [e-mail] wrote:
  

This makes sense to me, lets first identify the problems we want to solve 
and then see how to fix them rather than assigning NAT as a solution to 
problems that are not clearly defined.
    

Well, the main unsolved problem I see is

 - (small to medium) enterprise customers that want to change their ISP 
   without renumbering their internal network

For larger enterprises, I see "acquiring their own IPv6 address space"
(either by means of becoming a member of their local RIR and grabbing
a "provider" allocation, or by means of IPv6 PI space) "and using BGP"
as the answer to that problem.
  

The Stateless Address Mapping of draft-despres-sam-01 is intended to provide clean solutions to:
- SMEs that change ISPs
- Large entreprises that have multiple links to several ISPs.

The price to be paid is some complement if host stacks (backward compatibles for incremental deployment, but not available today). It is therefore not a solution for tomorrow morning, but it should not take that many years.

An experimental implementation should soon be planned, initially with Linux and  modified Linksys boxes,  and hopefully also with Wi-fi mobile phones and other hosts.

Pending availability of the solution in mainstream hosts, native IPv6 across IPv4 clouds, and IPv4 with its NATs and NAT cascades, should be sufficient.

I have native IPv6 today at home, thanks to the 6rd deployment by Free Telecom, with no plan to ever need NAT66s to do renumbering and efficient multihoming.

Regards,
RD