Re: [v6ops] IPv6 for mobile

[Rémi Després <[e-mail]>]

Le 25 déc. 2010 à 23:36, <[e-mail]> <[e-mail]> a écrit :
> ... 3GPP discussed the tools required for transition in length quite recently, and it was clear that

> the dual-stack and IPv6-only with NAT64 are the ways to go.

These two ways, however, shouldn't be believed to be on the same footing.
IMHO, people should understand that NAT64 is NOT as good as dual-stack for all applications of mobile devices.
 

> There are some discussions alive relating to host based IPv4 on-demand and double translation schemes, but

>  host based tunneling approaches have been clearly abandoned due complexities they bring. 

The lack of complexity of 6rd must have been either ignored or misunderstood.

> The 6RD does not really help in 3GPP based cellular networks, because the root cause is not lack of technologies to offer IPv6 connectivity,

Existence of several reasons for the slow deployment of IPv6 isn't sufficient to believe that 6rd can't accelerate it.

> So my opinion is that the lack of IPv6 here and there is best fixed by increasing the thing called demand.

Rapidly offering a *non disruptive* IPv6 service at a greatly minimized cost will increase IPv6 actual use by ordinary users (e.g. with Google).
New applications and user demand will follow.

BTW, do we know applications that can work across NAT64 without working also across cascaded NAT44s?

> ... Hence I would call for some patience; the existing transition mechanism toolbox is quite good but requires implementation, which will arrive when the time is ripe. 

My understanding of Tele2's need expressed by Mikael Abrahamsson on this thread is IPv6 connectivity:
- asap
- plug and play 
- without losing any currently available connectivity 

> Actually, I would personally be quite surprised if there were significant players in mobile industry still choosing to totally ignore IPv6.

So would I.

> The pipelines just have their variable lengths.

But if, at least for some operators, if this length can be reduced by a simple shortcut (based on a technology that has been validated on fixed networks), then why not?


Regards,
RD