Re: I-D Action: draft-ietf-6man-maxra-02.txt

[Lorenzo Colitti <[e-mail]>]

On Fri, Mar 10, 2017 at 1:12 PM, <[e-mail]> wrote:

>    The neighbor discovery protocol specifies the maximum time allowed
>    between sending unsolicited multicast Router Advertisements from a
>    router interface as well as the maximum router lifetime.  It also
>    allows the limits to be overridden by link-layer specific documents.
>    This document allows for overriding these values on a per-link basis.
>

The changes proposed by this draft are fine, but I have concerns about the
stated rationale.

Rationale: it makes no sense to say that we should be able to extend the
maximum RA timer above 9000 seconds because multicast is inefficient on
some links, and then cite a draft that says that multicast is not efficient
on 802.11 networks. On 802.11 networks, multicast is not as reliable as
unicast, but it's certainly not the case that one packet every 9000 seconds
is a problem. Maybe instead of citing 802.11, remove all mention of 802.11
and focus only on the 3GPP use case? Or, in addition to that, say that for
some power-constrained nodes, one packet every 9000 seconds might still be
too much.

Also: since we continue to see implementers do bad things such as set
maxinterval to 4s and router lifetime 12s on a wifi network, I would add to
the end of section 4:

As explained in Section 3, the relationship between MaxRtrAdvInterval and
AdvDefaultLifetime must be chosen to take into account the probability of
packet loss.