On Oct 14, 2009, at 20:45, Zach Shelby wrote:
I agree that within the LoWPAN some kind of 1-hop NUD may be useful,
and that should be discussed.
A couple of thoughts about that:
For route-over (where IP hop = MAC hop), much of that need is filled
by 802.15.4 ACKs.
(If there are no actual data, just a probe, it might be beneficial to
add a NOP dispatch type to 4944 at some point...)
One disadvantage of 802.15.4-2006 ACKs is that they are not secure.
15.4e is scheduled to fix this, but we may not want to wait for that.
This and mesh-under (where the ACK only covers the first mesh hop)
might motivate having some reachability probe message, but it would be
somewhat rarely used.
Using NS/NA for reachability is somewhat unfortunate, as these packets
are not small. Then, of course, if you want to derive arrival
probability estimations and have nothing else to go by, these are more
accurate if the packets are large.
The part I don't like about carrying over NS/NA just for NUD is that
the processing rules attached to these are more complex than needed
(cf. the ROLL discussion), and they can carry options, exacerbating
this.
With 4861 processing rules, many UDP-based applications would
regularly trigger NUD exchanges (as often as once per packet).
I don't think we want these when we have MAC-layer ACKs.
I don't like the last sentence of section 7 of 4861 here, even though
I understand that its desire for robustness is a reflection of some
real-world scars.
Gruesse, Carsten
PS.: Yes, I changed the subject line for tomorrow to be just about NUD.
_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan
