On Fri, 11 Feb 2005 09:09:57 +0200 (EET), Pekka Savola
<pekkas(_at_)netcore(_dot_)fi> wrote:
Yes, it was clear that this WG would only deal with IPv6.
It was not clear *why* this WG is being chartered to only deal with
IPv6.
What I fear is that unless the IETF does v4, someone else (ITU?)
will...
Again, we're not saying the IETF is not doing this. We're saying this
particular WG is not (cuz it's focusing on IPv6).
In any case, this could be clarified in the charter with a short
paragraph or a couple of sentences stating that there has been very
little interest in IPv4 adaptation so far, thus only addressing v6.
Ok. yes, it does make sense to document it. Technically, it makes less
sense than v6 cuz whereas with the current proposal we have the v6
header in common packets down to 2 octets (and it's straightforward
to reduce that to only 1 octet), it's not as easy to see this happening
with the more unpredictable IPv4 header.
Where was the determination made that adhoc routing and AODV in particular
is
the best fit in these scenarios?
The papers and the literature also abound (you can look it up via google).
The most definitive answer is that *NO* single protocol is the best fit
for all
scenarios. From a practical and engineering point of view, AODV is known
to work (there's even open source TinyOS code), and is also the choice taken
by ZigBee (so it makes less sense to go with something completely
different).
Should some of these reasons be captured in a sentence or two?
Makes sense.
How is that relevant to the IPv6-over-layer2
adaptation?
Mesh topologies are expected to be common in 15.4 networks.
However, a default mesh algorithm was not specified by the 802.15.4
spec. The idea is that whoever layers itself directly on top of 15.4 will
provide that. Zigbee has done so. If we are to layer IPv6 on top of 15.4,
then going the extra step and saying (use this mesh as a default
algorithm) is expected by the community who would potentially deploy such
devices.
Would AODV be extended to route based on layer2 addressing?
Exactly, it would exchange routing information about the 15.4 addresses
which are IEEE 64 bit by default (though 16 bit are also possible if doled
out by a coordinator), just as it has already exchange 128 bit addresses
(AODV for IPv6) and 16 bit addresses (AODV for TinyOS).
This
seems to go out of scope of the main v6 over Lowpan spec.
Above I explained why it is expected to be done by the WG. Not saying
this particular spec is the right place or not, though.
I can live with that. I just think the IPv6 over Lowpan spec is the
wrong place to do this. Maybe add another deliverable on Applying
AODV in (IPv6) Lowpan networks?
Perhaps, yes, gotta see how the WG review goes.
thanks,
-gabriel
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf