Jinmei,
> 2. we'd also like to allow users to use the /64 prefix on the p2p link
> just like an ethernet (i.e. broadcast) link. That is, when P/64 is
> assigned to the link, users can assume there might be a node P:A on
> the other end of the link for an arbitrary interface ID 'A', which
> might be learned from DNS, off-line communication, some other
> discovery mechanism, or even some misconfiguration or attacks.
Doesn't this mean that you'd like the other end of the link have more than
one interface ID? For the same reasons that folks use multiple interface IDs
on a multi access interface it would make sense to allow that on pt-pt
interfaces.
If you allow this there is a possibility that the other end uses a very
large number of interface IDs. Do you expect the two ends to somehow
discover (manually?) all the interface IDs?
Ignoring the DoS issue for a moment, there seems to be some utility
in sending anything in the /64 (except perhaps the sender's IP addresses -
different implementations seem to do that differently) across the pt-pt
link since it allows the peer to have any number of interface IDs in the /64.
*If* we want the above flexibility I don't think a simple
sending rule can fix the DoS issue - I think we need something akin
to sending an NS/NA pair across the link (even though that
sounds a bit silly on a pt-pt link).
Erik
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
