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

Re: New Version Notification for draft-koodli-ipv6-in-mobile-networks-02

Le 19 avr. 2010 à 19:30, Cameron Byrne a écrit :

>>> "In summary, IPv6-only deployments should be encouraged while considering the roaming and applications issues"
>> 
>> I still can't agree on this because:
>> - The market of real IPv6-ony is too limited.
>> - IPv6-only + DNS64-NAT64 is harmful to IPv6 e2e transparency.
> 
> 
> Why must IPv6 be pure? There is no precedent for IPv6 by nature of its
> structure being pure or meaning E2E transparency.


> NAT-PT has been
> around in the IPv6 ecosystem for a long time,

... but deprecated almost as soon as documented,

> so IPv6 addresses
> representing IPv4 destinations is by no means new.  

Not new as an idea, but AKAIK not deployed as a standard service.
Note that the DS-lite approach, in which IPv6-only access networks maintain IPv4 connectivity with NAT44s rather than NAT64s, doesn't need IPv4-embedded IPv6 addresses. 

> I agree that IPv6
> is the path to E2E transparency,

More than that: users of IPv6 today HAVE e2e transparency in IPv6.
DNS64-NAT64 in operator infrastructures, if done without serious precautions, is a path to BREAKING IPv6 e2e transparency.
This may be regrettable, but this in my understanding an inescapable fact.  

> but IPv6 does not imply e2e by virtue
> of the technology.

It just RESTORES e2e transparency, which had been lost when one address per host became impossible.

This happened long ago, and launched the painful complexity of NATs (cone, symmetric, port-resticted symmetric etc.), of their ALGs, of sophisticated tools to make pinholes in traversed NAT like TURN, STUN, Teredo, etc.


> IMHO IPv6-only expedites that journey to e2e.

Again, hosts that communicate today with public IPv6 addresses HAVE e2e transparency. 
This journey is therefore already finished... unless a step backward is consciously made to reintroduce complexity where it shouldn't exist :-(.

Note that NAT64 between a private scope IPv6 address (e.g. a ULA of RFC 4193) and IPv4, doesn't suffer from breaking e2e transparency, because private addresses have to be translated somewhere. This could provide a way for applications to know whether they can rely on e2e transparency or not, but this requires in my understanding more work.
   
Regards,
RD