Philip Homburg wrote: > > In your letter dated Fri, 22 Oct 2010 10:25:45 -0700 you wrote: > >Philip Homburg wrote: > >> In your letter dated Fri, 22 Oct 2010 11:05:42 -0400 you wrote: > >> I wonder what to make of that. If the SEND protected RS messages can > >> be replaced with AN-initiated (unprotected) RS messages, then what > >> purpose does protecting those messages serve in the SEND framework? > > > > The customer host will receive a SEND protected RA, which makes it > > possible to validate that it comes from a legitimate router (via > > certificates validation) and is not being replayed (via timestamps.) > > This implies that the end-device has to be able to match RS messages > using timestamp, i.e. its clock has to be sufficiantly accurate (to within > 5 minutes, according to the SEND RFC) to do that or (in the case of > failure) you would get hard to diagnose problems.
That would be a failure of SEND that is orthogonal to the presence of the mechanism described in <draft-krishnan-6man-rs-mark-08.txt>. > An end-device that requires its own nonce would fail similarly. If the end-device wants to receive a solicited RA containing "its own nonce", it sends an RS with said nonce, the AN encapsulates the RS up to the edge router, the edge router replies with an RA containing the nonce. > I think the draft needs more text about the interaction with SEND in > the case of failure. I disagree -- I do not see what in the mechanism described in <draft-krishnan-6man-rs-mark-08.txt> affects SEND failure modes. --julien -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
