Re: [6lowpan] Fwd: New Version Notificationfordraft-chakrabarti-6lowpan-ipv6-nd-simple-00

["Pascal Thubert (pthubert)" <[e-mail]>]

Hi Mathilde:

That's a great question that we need to bring to ROLL to be refined. In
short, RPL expects at least that the host is capable to notify its
router that it has an address that should be injected in the subnet
routing. RPL also expects that the host can tag a packet in such a
fashion that this can be interpreted by the router as an instanceID. The
details of how those things are done is not specified so far.

I would not favor a method that makes the host to router interface
specific to RPL if we can avoid that.

Best wishes for the interop!

Pascal


> -----Original Message-----
> From: Mathilde Durvy (mdurvy)
> Sent: Tuesday, March 23, 2010 7:42 AM
> To: Pascal Thubert (pthubert); 'Erik Nordmark'; '6lowpan'
> Subject: RE: [6lowpan] Fwd: New Version
Notificationfordraft-chakrabarti-
> 6lowpan-ipv6-nd-simple-00
> 
> Hi Pascal,
> 
> Thanks for your answer.
> 
> > - Section 7.3: I feel like you are underestimating the role that the
> > routing protocol might play. If you take the example of what RPL is
> > defining, an
> > IPv6 host could very well be a RPL leaf node, in which case it might
> > discover its default router by listening to DIO messages, and send
DAO
> > to 'register'. In addition, if the 6LR are RPL routers that use
DHCPv6
> > or another scheme for address allocation, RS/RA might be completely
> disabled.
> [Pascal]
> The classical way is that the routing protocol operates between
routers. ND
> provides the abstraction for a host to locate and interact with its
router.
> This is why in the current WG doc, the registration also belongs to
ND.
> [Mathilde]
> What you are saying makes sense. Although in 4861 the host is not
really
> registering with the router it is just discovering it. Hence, it is a
host
> -> router relationship but the opposite is not quite true, the router
does
> not use ND to route to the host (at least for off-link prefixes)...
> 
> [Pascal]
> Based on the registration, it is up to the router to redistribute the
> information in the route-over protocol, whether that is RPL or
something
> else. If we lose the ND registration capability, we end up forcing
every
> host attached to a RPL network  to support RPL. Don't you feel that's
wrong?
> [Mathilde]
> At this stage I don't know if this is right or wrong, but I would like
to
> understand what RPL is assuming? It's not so clear in the draft...
Maybe it
> relates to the questions that were asked on the role of leaf nodes
during
> the WG meeting yesterday.
> 
> Best,
> Mathilde