> I must chime in to say I dislike this approach a bit, but > it could be > salvageable. (Though, I must admit I like an option in RA's more..) > > Most of all, I dislike using *3* well known site local > addresses. One > should suffice just fine. Remember that this mechanism, at > least in my > eyes, is intended for *bootstrapping* only.
=> Hmm. I wonder if 1 address is enough. What if _that_ DNS crashed? Of course we can reuse that same address on other DNSs... oops we can't use anycast. I didn't really get the impression about this mechanism is only for bootstrapping, although the example you mention can certainly be done. Maybe the authors or someone else can clarify. After there is > something that > works, one could use some other mechanism (e.g. a DNS lookup :-) to > discover *real* DNS server addresses and ignore the site > local. It seems > to me that _3_ addresses just triples delays (5-10 seconds > each, per DNS > lookup?) if none of those have been configured in the site. => I think you're arguing for "how a host knows if it should use this mechanism" or "what is the deafult?". If this question is handled, the number of addresses is only relevant for redundancy. Hesham -------------------------------------------------------------------- IETF IPng Working Group Mailing List IPng Home Page: http://playground.sun.com/ipng FTP archive: ftp://playground.sun.com/pub/ipng Direct all administrative requests to [EMAIL PROTECTED] --------------------------------------------------------------------