Hi Remi, please see inlines ------------------ Peng Wu PhD candidate Department of Computer Science & Technology Tsinghua University, Beijing, China
------------------------------------------------------------- From：R閙i_Despr閟 Date：2011-04-06 20:57:34 To：Yong Cui CC：[email protected] list; Lee,Yiu Subject：Re: [Softwires] Questions on: draft-cui-softwire-host-4over6 >> You are right, Mark. So in order to encourage the deployment of IPv6 >> and >> make the deployment as simple as possible, we are looking forword to the >> 4over6 solution in CERNET2. > >With the customizable service you want, this is clearly a solution. > >But this of course doesn't prohibit others providers to prefer more >standardized services (for residual IPv4 connectivity across IPv6-only) to >privilege ease of operation, scalability, and investment costs. Depends on what kind of services a provider really wants. The other day at v6ops session this presenter from China Telecom said that the want their IPv4 & IPv6 addressing & routing be separated when provider this kind of IPv4-over-IPv6 service. So, it's still a tradeoff thing rather than standardized/not standardized... >The tradeoff could then be summarized as: > +----------------------------+--------------------+-----------------+ > | | Stateful (4over6) | Stateless (4rd) | > +----------------------------+--------------------+-----------------+ > | Customizable services | + | - | > | -------------------------- | ------------------ | --------------- | > | Ease of operation | - | + | > | -------------------------- | ------------------ | --------------- | > | Scalability | - | + | > | -------------------------- | ------------------ | --------------- | > | Invetment cost | - | + | > | minimization | | | > +----------------------------+--------------------+-----------------+ Well, if we talk about these kinds of pros & cons, keep in mind that 4over6 is an extension of Dual-stack lite, they can be naturally coupled(shared devices & services). So I think the "-" items doesnt't match. IMHO, theoretically the relationship is like this: if don't like address & routing coupling, then stateful(on AFTR). Under this direction, if dedicated v4 address, then 4over6; if stateful address sharing, then DS-lite; if stateless sharing, then some sort of A+P thing; if can accept address & routing coupling, then stateless. if dedicated v4 address, then 4rd; if address sharing, then 4rd with address multiplexing _______________________________________________ Softwires mailing list [email protected] https://www.ietf.org/mailman/listinfo/softwires
