Ralph Droms <rdroms.i...@gmail.com> wrote:
>>> But with vehicles, one connects a vehicle here and gets a prefix, then
>>> moves in that area and gets another prefix. At that point, if the
>>> router obtaining a prefix wants to delegate further to another vehicle
>>> needs to change the delegated prefix.
>>
>> Why wouldn't RPL be used for such networks? It has built-in PD for
>> dynamic networks, if I understand it correctly, with RA used at the
>> subnet level.
RD> My understanding is that RPL assumes a multi-link subnet, with a
RD> single prefix used across the entire network and source routing or
RD> host-specific routing.
RPL doesn't care.
It's all /128 routes to anything not on the local (physical) link.
Whether it's source routes or hop-by-hop LPM depends upon whether the
mode of operation is storing or not.
They don't have to come from the same subnet. The RPL Target Option,
contained in the DAO that travels up the tree, has a prefix and
length. In typical layer-3 subnet mesh networks, it contains /128
prefixes.
Like others on this thread, I'd like to know how real this vehicle use
case is. Is CAE really working on such a thing?
My opinion is that vendor of vehicles should obtain a Non-Connected
Network prefix from a suitable registry, or from
not-yet-agreed-to-be-useful ULA-Central, and have the the MR-IV
and MR-LV (permanently) number the LFNs.
When two cars connect for some non-Internet access reason (including to
talk into the homenet), they should speak RPL to each other across the
E1 link(s). The DAG would not be grounded.
Like all the other home LLNs, the gateway into the homenet would need to run
RPL on one side, and homenet-routing-protocol on the other side (zOSPF,
whatever). It would announce the prefixes of the vehicle into the
homenet (another kind of "walled garden"!!), just like the lighting
system does.
If Internet connectivity is available/desired, then one of two things
occur (not both):
1) the Internet connected vehicle obtains topologically significant
(PA) address via homenet-routing-protocol, and then forms a new
(grounded) DODAG which can be shared with adjacent vehicles that
want connectivity. This solution is probably more secure and
probably will work in many non-homenet environments, since
homenet-routing-protocol isn't the only way to get prefixes, and
even if you get only a single /64, RPL will let you "share" it.
2) the Internet connected vehicle becomes a homenet router, and
speaks homenet-routing-protocol on E1 link and enough address space
for all vehicles is obtained.
The PA addresses live only for the duration that the vehicle(s) are
connected.
Again, maybe I missed the bigger picture: who is building these IPv6 enabled
vehicles? (And where can I get one?)
--
Michael Richardson
-on the road-
pgpuQE94cXXdj.pgp
Description: PGP signature
-------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
