Re: draft-arkko-ipv6-transition-guidelines WGLC

[Cameron Byrne <cb.list6@gmail.com>]

On Fri, Aug 20, 2010 at 1:41 PM, Fred Baker <fred@cisco.com> wrote:
>
> On Aug 20, 2010, at 12:10 PM, Cameron Byrne wrote:
>
>> IPv6-only host must emerge as the edge grows (mobile ..., M2M, ...)
>
> I certainly agree with that. Understand that mobile handsets are already and increasingly both IPv4 and IPv6 capable. The issue is in the network, and it is in the applications that those handsets use. The applications like to call gethostbyname() instead of getaddrinfo(), and as a result are IPv4-aware instead of being network layer agnostic. That was Hui Deng's point in the two 3GPP/IETF joint workshops and in the pnat discussion in behave. This document grew out of a powerpoint presentation, pointed to as an informative reference, that I gave at the first of those workshops, and summarizes both that talk and Jari's experience using Ericsson's NAT64 solution at home.
>

Just focusing on mobile, here ... Jari states IPv6-only works well for
mobile.  I believe one of his docs (Google preso)  states it
approaches 100% functionality.

PNAT did not achieve much support and was not advanced (right?),
despite a strong effort from Hui.

Like IPv4 literals, i believe the general issues in mobile apps are
trivial.  I presented my experience at the Google conference.  Notice
not many apps are impacted
https://sites.google.com/site/ipv6implementors/2010/agenda/13_Byrne_T-Mobile_IPv6GoogleMeeting.pdf

My FUT users, along with my own experience, is that very few apps fail
when ipv6-only

http://groups.google.com/group/tmoipv6beta/browse_thread/thread/6fc0633f0313b588

I understand that the desktop space, and in particular games and
skype, have problems with IPv6-only.  But, mobile is different.  The
SDKs for Apple, Symbian, and Android have done a good job at forcing
the developers to be IP version agnostic..... which means most apps
are IPv6 capable by default, not by app designers own volition.


> If I were king, and note that I am not at this instant wearing my shiny working group chair crown, I'm wearing my dusty author's beret, gethostbyname would disappear and applications that use it would be given the option of complaining to /dev/null or fixing their code. At some point, maybe the 3G vendors could help us out there...

As far as i can tell, for the most part, your desire has been
achieved.  If you believe otherwise, please contact me off list to
find out how you can join the IPv6-only beta.

Cameron