> So could someone with the full overview outline explain to me
> how many mechanisms a RPL node running over 6lowPan will have
> to implement to be compatible with all other nodes claiming
> to be RPL compliant?
> 
> My guess:
> * Classic RS/RA
> * DHCPv6
> * 6lowPAN ND
> * RPL address assignment
> * etc
> 
> Should we make a decision someday? (!)

Anders,

that is indeed a very important point.
We need to get rid of as many of these options as possible.
At least for the baseline expectation.

In the 6LoWPAN context, we need to distinguish three kinds of nodes:

1 -- hosts, which should have absolutely minimal requirements imposed on them;
2 -- router nodes (6LR in ND-09 terminology), which need to be more complex to 
do routing, so they probably can also implement some additional complexity for 
other purposes such as bootstrapping;
3 -- edge routers (6LBR in ND-09 terminology), which have multiple interfaces 
and are more likely to be relatively powerful.

1) Hosts don't participate in RPL, so there cannot be any requirement from RPL 
on them.
Of course, 6LoWPAN-ND must do everything to enable a host node to participate, 
and to enable the router node(s) to represent the host in the network (6LoWPAN 
and/or RPL LLN).

2) Router nodes participate in a routing protocol.  Since RPL does not seem to 
address mixed networks, from an RPL point of view it can be assumed that they 
participate in RPL.  But they have only one interface, so that participation 
can still be governed by 6LoWPAN-based assumptions.

3) Edge routers have multiple interfaces, more than one of which might be 
engaged in RPL.  Full complexity, likely including DHCPv6 prefix delegation etc.

4) From the point of view of RPL, there is a fourth kind of node: One that has 
no 6LoWPAN interface.  If that is not using 6LoWPAN-ND, it needs other ways to 
achieve bootstrapping.  I would not be happy if RPL would grow features that 
are needed only for bootstrapping this fourth kind of node and still make them 
mandatory for type 2 nodes (or even try to browbeat type 1 nodes into using 
them!).

Gruesse, Carsten

_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan

Reply via email to