>Erik's draft doesn't require a two-faced DNS. If I recall correctly, you
>simply don't use the site-local addresses unless you recognize the global
>prefix as being in your site (which you know from site-prefix
>advertisements).
if i understand correctly, Erik's draft requires (or a "should")
every resolver to implement Eric's way of identifying local/remote
sitelocal peers on DNS. if a naive client implementation (without
support for Eric's draft) queries DNS and got site-local address
from remote, the client will try to contact the address in its own
site, to get no response. I understand that Eric's draft is an
additional MUST to IPv6 nodes, if we want it to get deployed.
i'm not really sure if we can do it at this stage.
at least BIND4/8/9(lwres) do not seem to implement it.
it is rather unlucky, but two-faced DNS server is easier to deploy
(since it only impacts server side, not resolver side)
(to clarify: it matters only if you want to resolve site-locals in
your site by DNS. if you just want site-locals for admin purposes
like router renumber, you don't really need to populate site-locals
in your DNS database)
itojun
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
