> It seems that you have missed some statements in the draft. In section 4.3, > there are some lines clarifying 3 principles for IPv4 interface selection, > just > above the words you quoted: > >1. IPv6 transport is preferred over any other. > > >2. Less address translation occurrences is preferred over more. > [RFC5864][I-D.donley-nat444-impacts] > > >3. The closer the state is to the edge, the better.[RFC1958] > > According to the first principle, the IPv6 transport is prior to any other. > That > is the reason why the DS-Lite tunnel interface is preferred to native IPv4 > one. > > Actually, I have another question about this point. I think the question is > not 'Why not use Native IPv4 interface for better performance', but 'Why > prefer > the IPv6 tunnel interface to the native IPv4 interface'. I didn't find the > answer in > the draft. Perhaps I've missed some words as well. IMHO, a clarification of > the > reason could be provided in the draft, and thus the principles about might > seem > to be more convictive.
the idea was to allow for a controlled move from a native IPv4 connection (that is being sunet for an IPv6 only access) to IPv4 over DS-lite. the policy table proposed is all about IPv4 exit, not trying to optimize what IPv4 link-layer may give the best IPv4 service. cheers, Ole _______________________________________________ Softwires mailing list [email protected] https://www.ietf.org/mailman/listinfo/softwires
