Brian E Carpenter a écrit :
On 2008-01-29 22:02, Rémi Després wrote:While it identifies some threats in some configurations, it doesn't conflict, in my undersatnding, with mapped addresses on the wire in some other useful ones. One of these other configurations happens to be very pertinent (and IMHO completely Kiss :-) ) for IPv6-only client hosts to be able to reach IPv4-only servers.There's no fundamental need for a magic prefix. Iljitsch and I will have a combined draft out soon that will discuss this. Being able to use a regular unicast or anycast prefix to reach the translator is more KISS than a magic prefix, IMHO.
I agree that other prefixes than mapped address prefix (0::/64) can be proposed to go across a v6 cloud to the v4 world. But I see them as an indirect solution to a need that can be satisfied with a direct one.
This "unicast OR anycast prefix" needs being invented, and specified, while the mapped address prefix exists already, and to my knowledge works nicely.
With the mapped address prefix, IPv6-only hosts that send packets to IPv4-only hosts have, without any address translation, the right routing prefix in their destination address. They are *real IPv6-only*.
It seems to me that the combined draft you work on would be significantly simplified (and therefore improved) if it would replace the "unicast OR anycast prefix" by the existing mapped address /64 prefix. (I believe, like you and Iljitsch do, that scalable NATs between v6 and v4 are a useful approach for the v4-v6 coexistence period, and that for this they may have to be split. But I also believe that some improvements are still possible in this approach. Accepting and using the mapped address prefix would IMHO be one of them.)
Maybe I miss something, but the question is "what is it exactly?". Rémi