Re: draft-bourbaki-6man-classless-ipv6-00

[Simon Hobson <[e-mail]>]

Lorenzo Colitti <[e-mail]> wrote:

> When all the applications have built NAT traversal mechanisms, anything more than one IP address per location becomes worth nothing. At that point the extra rent is gone, and everybody loses because the system is less capable, less robust, and harder to configure than it would have been without NAT.
> 
> And in fact, that's what happens in IPv4 today.

Agreed - but that is typically out of necessity as there just aren't the IP addresses to go round.

Having had the luxury of managing a network with a whole /24 to play with, and looked at the underlying problems making SIP phones work through NAT, I fully agree that NAT is to be avoided due to the amount of brokenness* it creates. Trying to persuade those who see things "just work" (because of all the wasted effort put in to make things work - would have been much better spent getting IPv6 adopted sooner) is a different matter.

BUT, I cannot see the logic behind your earlier comment that DHCP makes NAT inevitable for IPv6.


* PS - I have a special place in hell for the ******** ******** at Zyxel who think that randomising port translations on every connection, and "no you can't turn that off, we don't care if it breaks things, it adds security" (yes, that's the response I got from them !) is a good idea.