On Tue, 17 Aug 2010, Fernando Gont wrote:
AFAICT, it does. It says: "....and the destination address on the packet
seems to be on-link (in terms of Neighbor Discovery) on the
point-to-point interface". Or am I missing something?
Yes, you're right. I was not reading the text carefully enough.
(Though "is on-link" vs "is of the same subnet prefix" are
semantically two subtly different checks. Not sure if it matters in
this case.)
It seems that the point is not really that of reduced performance, but
rather that complying with this requirement would require a change in
the silicon?
If that's the case (i.e., no real performance implications), then it
looks like an appropriate fix for this issue. -- which does not
necessarily argue against /127 prefixes, as there are other reasons for
using them (or, put another way, let's not correlate *this* with the
fight over /127 prefixes).
This issue was initially brought up by Google IPv6 presentation,
proxying Juniper's statements, so it would probably best if either of
them could clarify.
FWIW, "Packet may be forwarded back on the received interface" is
actually, AFAIK, used in certain PE routerscenarios where you ping
yourself over a p2p link.
Is the echo request/response really forwarded back on the received
interface? (isn't the *response* that is forwarded back on the received
interface?)
You're probably right. I was likely confused on how this is actually
done, and right now I can't find any written references on the "p2p
self ping" troubleshooting technique I seem to recall. Olivier's note
about the different scenario may still apply.
--
Pekka Savola "You each name yourselves king, yet the
Netcore Oy kingdom bleeds."
Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
