Re: [BEHAVE] RFC6147 and RFC7208 interoperability issues

[Keith Moore <[e-mail]>]

Sent from my iPhone

> On Feb 7, 2022, at 1:40 AM, marcelo bagnulo braun <[e-mail]> wrote:
> 
> My guess is that NAT was done wrong because the IETF decided against standardizing NATS in due time because it was a kludge that should not be done, maybe?

No I don’t think so.  I don’t think anyone ever figured out how to do NATs even close to “right”.  I tried to do so for many years, and never managed to solve some of the serious problems.  And of course by that time “wrong” NATs were so widely deployed that there was no hope of fixing them anyway.

Bottom line: NATs are inherently incompatible with the internet architecture and with any network architecture that offers near-complete interconnection between hosts.  And also inherently incompatible with DNS. 

The only reason that NATs have been able to survive is that they more-or-less worked with the web and email.  But they crippled many other applications.  Meanwhile email has since (to my great disappointment) been crippled in other ways, and the web is a security and privacy disaster, and it turns out that people need a wider variety of applications than the web and email anyway.  


> 
> NATs are reality (I hope we all agree with that). I guess it is more productive to try to provide guidance about how to make them as harmfull as possible rather that ignore them, provide no guidance and end up with all possible version of NATs, including the worst possible ones.

Saying that something is reality is no justification for keeping it.  I suppose it would be more precise to say that NATs have mindshare - lots of people think that NATs are just how the internet works.  But the problems with NATs and also with IPv4 are more widely understood than ever.  It’s time to abandon both and set an EOL date for the public IPv4 internet. 

Keith