Re: New I-D on Carrier Grade NAT

[EricLKlein@softhome.net]

Tomohiro Nishitani writes: 

> Hi, 
>
> I submitted new draft on Carrier Grade NAT. 
>
> "Carrier Grade Network Address Translator (NAT) Behavioral Requirements
> for Unicast UDP, TCP and ICMP" 
>
> http://www.ietf.org/internet-drafts/draft-nishitani-cgn-00.txt 
>
> Abstract of our draft is as follows: 
>
> ---
> This document defines basic terminology for describing different
> types of carrier-grade Network Address Translation (NAT) behavior
> when handling Unicast UDP, TCP and ICMP. Developing carrier-grade
> NATs that meet this set of requirements increase transparency of data
> between carrier networks.
> ---

As much as I am all for killing NAT as soon as we can, I understand the 
potential need for such a device as currently carriers are doing just this 
using multiple devices (and frequently nesting NATs). 

I would prefer that the Abstract was changed to the following: 

This document defines basic terminology for describing different\types of 
carrier-grade IPv4 Network Address Translation (NAT) behavior when handling 
Unicast UDP, TCP and ICMP. Developing carrier-grade NATs that meet this set 
of requirements increase transparency of data between carrier networks. 

Where there is an emphasis on the fact that this is IPv4 only, not a 
backdoor for bringing NAT into v6. 

I also think you need to consider the requirements for non-stationary 
devices as these are starting to be more common and will complicate CGN to 
no end as they move from one cell site to another or into a new carrier 
while trying to maintain their IP session (most critical if it is UDP 
based). 

Eric 

Eric