Richard,
good points. It appears to me (maybe others) that some of the basic
assumptions: mesh under or star networks are edge cases, that meshes
only use broadcast flooding, that DAD is necessary, and that RA/RS can't
be made to work on mesh under and route over network, that anything more
than RA/RS are necessary are flawed.
I agree that finding some optimizations for 4861 would be good. Erik and
Samita submitted an ID that had some suggestions for ND optimizations
not replacement.
I'm not convinced that we need the complexity of white boards, new ICMP
messages, more messages being transmitted to keep state alive, turning a
stateless ND protocol into a stateful one.
If 4861 won't work as is in a Route Over network (it does on mesh
under), then we need to find a way to fix it, but that doesn't mean we
have to go to the extreme of inventing something new (whiteboards,
nr/nc, 6AO, 6IO, 6SO, ...) and creating something that is incompatible
with existing infrastructure, devices, and management policies and
techniques.
geoff
On Mon, 2009-11-09 at 18:00 -0500, Richard Kelsey wrote:
> As a newcomer to IPv6 and ND, I am having trouble following
> the "4861 works"/"no it doesn't" debate. The ND draft lays
> out the reasons why its authors think that 4861 is unsuitable
> for low-power/lossy networks. Clearly, some folks disagree.
>
> Is there anything that explains where the authors of the ND
> draft went wrong? Are there descriptions of how 4861 has
> been used in this sort of network? I understand the
> downsides to not using 4861, but if it won't work, what
> choice do we have but to use something else?
>
> -Richard Kelsey
>
> _______________________________________________
> 6lowpan mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/6lowpan
_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan
