Itojun,
>
> when there's linklayer address defined for the particular L2,
> we don't do NS/NA (from the spec). please imagine ppp, or rfc1933
> tunnels. do you think it reasonable to run NS/NA in that case?
> then, what kind of changes are necessary to NS/NA?
>
I am imagining it. My point is, that maybe the restrictions you are placing
on the solutions space don't make sense given what you are trying to do.
That is possible, right?
The thing you are describing, a link with a /64 assigned to it, doesn't
look like a point-to-point link because one of the points is not specified.
It is a point-to-(maybe multiple points none of which are known) link.
This configuration kind of requires a way to discovery what is on the other
end of the link.
Since we already have a protocol for doing something like that discovery,
namely NS/NA exchanges, it seems reasonable to look at that first as a solution
rather than come up with some second order modification to forwarding rules
to deal with the downside (e.g. DoS attacks, wasted bandwidth because of
loops, etc.) of not knowing precisely what addresses are assigned to your
neighbor.
Sure it may require some textual changes to ND to allow NS and NA packets
that don't have SLA and TLA options but are the code changes that difficult
to imagine? Is there more to it than that? I am not positive but I don't
think so.
Your proposal is the expedient way to go but if a point-to-point link with
a /64 assigned to is a useful thing then I would prefer to see it supported
in the protocol in a first class way.
Tim Hartrick
Mentat Inc.
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
