Jonathan,
(no WG chair hat)
Apart from the philosophical argument made by Pascal, there is a big
difference in complexity between 6LoWPAN-ND's NR and DHCP.
Your semantics point ("stateless"):
SAA needs DAD. 6LoWPAN-ND does DAD in a different way, because it
cannot rely on transitive multicast. But it still does SAA.
If you really want to focus on the "stateless" in SAA, what about the
states that an SAA node is in (optimistic, tentative, preferred,
deprecated). I can already argue that original SAA is not "stateless"
for this definition of stateless. So I'm not surprised you can do
that for your favorite definition.
The 16-bit address search (it's not really assignment) goes beyond
SAA, yes.
It appears to be a good optimization (and that's what 6LoWPAN-ND is
about) to not let the node guess repeatedly until it finds a free slot.
I don't see an additional server, the ER is needed in any case.
The ER better knows which nodes are in the LoWPAN (or it will create
lots of pointless RREQs), so it is absolutely natural for it to have
that table ("whiteboard").
Gruesse, Carsten
_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan
