Reviewer: Jouni Korhonen
Review result: Ready with Nits
I think the document is close to ready but some more meat in still
needed. Similar type of configurations have been used in other types
of link with success. A couple of nits and questions.
Line 157: "hosts/subscribers devices connected to the provider managed
shared" is somewhat hard to parse. Does it mean to say
RFC6106 reference need to be replaced with 8106.
Line 259: Retransmit timer is the "AdvRetransTimer" or "restrans
timer"? Be more specific here.
Section 4 talks about unicast RAs. I think it would be good to mention
also RFC6085, which allows unicasting an RA to a host on an Ethernet
link, while RAs still use layer-3 multicast addressing. Similar goal,
different approach. At least I would welcome text why RFC6085 is not
endorsed or applicable?
Section 5 gives some guidance regarding unsolicited RAs. However, it
does not say anything about PIOs, although the context of the text
seems to assume the presence of PIOs. I would mention those
explicitly. I would also be more clear and state the unsolicited RAs
are again unicasted. At least the current text seems to assume so.
What happens when RSes get lost (all MAX_RTR_SOLICITATIONS)?
How shared links with multiple first hop routers are supposed to work
with the recommendations of this BCP. Current text seems to be only
considering a single first hop router deployments.