On Oct 14, 2009, at 19:15, Mathilde Durvy (mdurvy) wrote:
[Mathilde] What about a router with two interfaces, one Ethernet and
one
802.15.4. With the current proposal this router will need two
implementation of ND, 4861 on the Ethernet interface and 6lowpan-ND on
the other.
As with any node that has interfaces into 4861-supported networks as
well as into networks not supported by 4861.
(Actually, such a Ethernet-to-LoWPAN router will in addition also need
to implement the edge router function of 6lowpan-ND, which intra-
LoWPAN nodes don't need to do -- we are trying to simplify the latter,
in preference to the [relatively!] powerful Ethernet devices.)
An important implication is that we need to modify the OS (where ND is
implemented) of exiting computers / routers / etc to allow them to
use a
802.15.4 interface.
As has been mentioned a couple of times, 6lowpan-ND can be (and has
been) implemented as an adaptation layer, satisfying any ND layer that
cannot be mapped out of the OS architecture and keeping it ignorant of
the actual nature of the link; since 6lowpan-ND only modifies/replaces
a couple of 4861's messages, that's almost a bit like header
compression.
I'll now stop repeating all this over and over, because it has been
mentioned often enough on the mailing list.
I still don't understand why we had to go through seven iterations of
the adopted WG draft before anyone noticed that, yes, we have a
version of ND that has been optimized for 6lowpan and is therefore no
longer the same as 4861.
Gruesse, Carsten
_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan
