[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: I-D ACTION:draft-bagnulo-v6ops-6man-nat64-pb-statement-00.txt



On 23 nov 2007, at 16:51, Norbert Bollow wrote:

I think I'm getting convinced that your "modified NAT-PT" approach
http://www.ietf.org/internet-drafts/draft-van-beijnum-modified-nat-pt-02.txt
is essentially the way to go.
My only concern is that there is not much time left until IPv4 address
depletion will start causing real problems, and we don't even have a
"finished" RFC for "modified NAT-PT" yet.
How do you see the timeline with regard to getting more feedback on
this I-D, publishing the RFC, implementation and deployment?
That's a good question. I haven't paid attention to any instances of  
the IETF moving fast with a protocol RFC, so I can pretty much only  
guess. But I'd say it would take about a year to reach consensus and a  
stable specification if there is sufficient interest, except that  
carving out a block of IPv4 space for the IPv4-to-IPv6 part could take  
more time.
As for implementations: the IPv4-to-IPv6 part requires some changes to  
existing translators. The IPv6-to-IPv4 part probably doesn't, but I'm  
not completely sure about that. On the (IPv6) host side, the IPv4-to- 
IPv6 part can work for the most part without changes. But the IPv6-to- 
IPv4 part requires changes fairly deep inside the TCP/IP stack, which  
will probably take a considerable amount of time to roll out on  
operating systems like Windows and Mac OS even if Microsoft and Apple  
give this some priority. The problem here is that you don't get much  
benefit until both translators are deployed in the network and  
operating systems are modified to make use of them.
Alternatively, individual ISPs could roll out translators in their  
network and provide customers with CPEs that perform IPv4-to-IPv6  
translation as outlined in section 5 of the draft. These CPEs only  
have to implement SIIT without the NAT part plus some address  
management, so implementing that wouldn't have to take forever. The  
fact that both ends are under the control of the same entity will  
certainly help get things off the ground faster.
I would guess it's reasonably possible (but not a given) to have  
something on the market in 2 years