On 2004-02-26, JINMEI Tatuya / [EMAIL PROTECTED]@C#:H wrote: > >>>>> On Thu, 26 Feb 2004 20:54:32 +1100, > >>>>> "Nick 'Sharkey' Moore" <[EMAIL PROTECTED]> said: > > > My only concern would be whether or not there needs to be > > a requirement to defend LINKLOCAL::SUFFIX when configuring > > UNICASTPREFIX::SUFFIX.
[...] > Sorry, I don't understand the concern...by which scenario might the > latter not be defending the link-local address? (perhaps I don't > understand what you meant by "defend"...) Sorry, that's me being obscure: By 'defend' I mean 'send an NA in response to an NS for that address from the unspecified address', eg: to 'defend' that address as being yours. The scenario being that under non-IID based allocations such as RFC3041 or SEND-CGA, a node offered prefixes A, B and C may configure different suffixes X, Y and Z ... thus the addresses A::X, B::Y, C::Z. This breaks the old assumption that all allocated addresses (on a given interface) will have the same suffix. In order to defend the address C::Z from 'old-style DIID' nodes, the new-style node would have to reply to NSs for LINKLOCAL::Z too. I'm not sure that's explicitly explained in the text (maybe I've just missed it?) cheers, hope that clarifies things? -----Nick -------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
