Re: [v4v6interim] "IPv4->IPv6 is hard"

On Oct 17, 2008, at 11:29 AM, Ed Jankiewicz wrote:

> yup.  I really meant to say "right now" and even for the near  
> future, don't we still expect most servers/applications to be dual- 
> stack rather than IPv6-only?  My point was there are more pressing  
> problems to solve, while it would be nice for all variations to  
> work, some can be deferred.

One has to define "dual stack".

Everything that runs on my Mac is potentially dual stack, in the sense  
that the code is there to run both IPv4 and IPv6. Present it with an  
IPv4 network, it does so. Present it with an IPv6 network, it does so.

It the thing it needs to talk with is IPvX and it only has IPvY  
routing, however, it can't get there. From my home I can't access  
ipv6.google.com because no ISP in Santa Barbara County runs IPv6.

The DS-Lite model says that I should set up a tunnel across the IPv4  
infrastructure, whether from my host or from my router; 6to4 is  
similar, as are others.

The translation model says that the host should use the network  
available to it and something in the network should change the  
protocol when and where it must.

My personal preference is for dual stack deployment, and where  
necessary to change as little as possible, as the changes will outlive  
the need for them by decades. Asking the host to build tunnels is an  
example of such a change. Under the ISP's control or at least in  
collusion with it, I can see building tunnels. But even there, as the  
ISPs deploy IPv6 that becomes a change that will outlive the need for  
it. That's why I'm working on translation - enable the network  
administration to install what it needs to when it needs it, and  
enable it to remove it when the need goes away.

my two yen

_______________________________________________
v4v6interim mailing list
v4v6interim@ietf.org
https://www.ietf.org/mailman/listinfo/v4v6interim