Michael's assumptions regarding why are mostly correct. There are many versions 
of the SSNI network, but the standard model is where the continuously powered 
devices (CPDs) which are firmware-upgradable form the core canopy and the 
majority of the device population, and various battery-powered devices hang off 
of the canopy using whatever CPD signals they can get. The CPDs are typically 
not significantly constrained (they aren't constrained at all per rfc7228) but 
the BPDs typically are, and they aren't all the same rfc7228 class. The 
evolution is to allow standard Wi-SUN (route-over) devices join the mesh, and 
potentially use some of the same RPL data structures, messages, and algorithms 
where possible to reduce the private burden of having non-standard routing 
structures and the various tooling and expertise challenges that go with that.


Benjamin Damm
O: 669-770-4000
E: [email protected]  www.ssni.com


________________________________________
From: 6lo <[email protected]> on behalf of Michael Richardson 
<[email protected]>
Sent: Thursday, April 13, 2017 7:51 AM
To: Pascal Thubert (pthubert)
Cc: [email protected]
Subject: Re: [6lo] Adaption of ROLL for mesh-under

Pascal Thubert (pthubert) <[email protected]> wrote:
    mcr> I would suggest that networks that want to move from mesh-under to
    mcr> route-over and are already running IPv6, would be best to use RPL in
    mcr> the way 6tisch uses it.  Gather the topology with RPL, then
    mcr> establish  6tisch-like tracks at L2.

    > [Pascal] In a green field, yes; maybe leveraging DAO projection,
    > too. But then if the goal is to use an existing forwarding plane,
    > certainly that's not the same story.

yeah, I'm not entirely sure I know what the existing situation is.
My impression is that the idea is to have a network that is a hybrid of both
mechanisms, because the network is being incrementally evolved.

As such I can see perhaps making L3 adjacencies (parent/child) over multi-hop
L2 mechanisms,  but in some cases, also recognizing that there are L3 aware
nodes in between.

If the nodes capabilities are high (and software upgrades or incremental
replacement is possible), but the network bandwidth is such that running two
networks (on different PANIDs or something) is impossible, then this might
work.

I asked Benjamin for more details privately, and I think that we'll get some
more details soon.

--
Michael Richardson <[email protected]>, Sandelman Software Works
 -= IPv6 IoT consulting =-




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

Reply via email to