Re: [Roll] I-D Action: draft-ietf-roll-useofrplinfo-04.txt

[Michael Richardson <[e-mail]>]

Michael Richardson <[e-mail]> wrote:
    > Again: my point from before, we can get rid of hop-by-hop IPIP headers by
    > changing the Option Type, or by creating some signal that there
    > are no non-RPL nodes.

The revised proposal at 6man this morning (not yet in text of a draft) is
that all current Option Types become ignore if node does not understand.
(A note can be configured to understand a HbH and drop it)

This means that we do not need to change our Option Type to get the behaviour
that I think we would like.

It also means that we do not need to remove the RPI header before sending
traffic to the Internet.  This is significant because it means that we do not
need an extra IPIP header for traffic leaving the LLN.
We still need an IPIP (+RPI+ RH3-for-non-storing) for incoming traffic, and
we still want to compress it.

That incoming traffic *COULD* now include an RPI from another LLN, but that's
okay, we probably have to encapsulate in IPIP to add our own RPI.
Could we reuse an RPI that is already there, I'm not sure.

A really important result is that we can remove all cases and consideration
of cases where the target is not known to support RPI.

If this WG agrees with the 6man proposal, I will remove those cases from
useofrpi and explain why.

--
Michael Richardson <[e-mail]>, Sandelman Software Works
 -= IPv6 IoT consulting =-