Re: [v6ops] Worse than NATed IPv4? [was IPv6 for mobile]

[Rémi Després <[e-mail]>]

Le 25 déc. 2010 à 22:58, Gert Doering a écrit :
> ...
> Basically, there's three major groups of player here, and to blindly state
> that one solution is going to work for all three of them is too close-
> minded.

What would IMHO be close-minded would be not looking at "6rd across the IPv4 bearer service", proposed as a workable solution for these 3 player groups:  

1.
> - established Internet access customers (DSL, cable, ...) - these already
>   have IPv4, and could run dual-stack - but they will just stay with IPv4

>   unless their ISPs will offer some sort of incentive to move to IPv6
>   (or the ISP will just "make it happen" without the customer knowing)

With 6rd, the ISP can "make it happen" very easily (as Free did in production in 2007, and as Comcast did in one of its trials in 2010.) 

2.
> - new growth in Internet access customers - there's no IPv4 for those, so 
>   they can't run dual-stack without NAT444 for IPv4 - but we'll see those 
>   deployments anyway.  

o Agreed, there will be NAT44s in operator networks. (AFAIK, there are some already. For high-end handsets that act as routers this will indeed be NAT444 scenarios.)
o In PCs, NAT64 is more disruptive than NAT444 plus IPv6 (easily feasible with 6rd).


> A subset of that group is "mobile" - very controlled
>   device and program environment, and the network operator knows precisely
>   what sort of device is trying to use the network, so IPv6-only can be 
>   made work, handset vendor-by-vendor

o I agree that IPv6-only+NAT64 may work for specific devices whose applications are restricted, and under operator control (e.g. web browsing without transmission of addresses, and e-mail). For these devices, 6rd can work too, but isn't a must. 

3.
> - content - here we'll *have* to go to dual-stack, because there will be
>   IPv4-only access customers for a long while, and for the networks that
>   have no more IPv4, "direct IPv6" has less overhead and less middleboxes
>   than NAT64 - so has a higher propability of working

o 6rd offers a real dual-stack service. 
o the IPv4 header added to IPv6 packets won't be noticed in user normal operation. 


Regards,
RD