Hello Nicolas, Erik, > > > > As Pascal answered in his post, the TID is not particularly coupled > > with RPLv1. The TID is just an extra information provided by the node > > during ND registration which dramatically simplifies node localization > > but also enable DAD across a backbone of Edge Routers advertizing the > same prefix. > > Which alternative solution would you suggest for DAD? > > DAD works fine with what we have in ARO. > If the same EUI-64 (re)registers the same IPv6 address, then it is not a > duplicate. > If a different EUI-64 tries to registers an IPv6 address (already registered with > some other EUI-64), then it is a duplicate. > That is independent whether the check is done in a 6LR or 6LBR. > > What do you see as the problem with DAD? Can you provide an example of > what doesn't work with 6lowpan-nd?
[Pascal] TID is not related to uniqueness/DAD. It is related to movement. A node that stays attached to the same router would not need it. Tragically, LLNs are not like that. TID enables to invalidate states that have a deprecated sequence. Typically if you want to redistribute an ND registration into something else then you need such thing. I listed in another post 3 issues with ND that can be fixed with a TID (scalability, out-of-order, and anti-replay). DAD is not in the list. DAD works as long as the EUI is unique, which is an assumption we accepted for this work. Pascal _______________________________________________ 6lowpan mailing list [email protected] https://www.ietf.org/mailman/listinfo/6lowpan
