Re: [v6ops] New Version Notification for draft-baker-v6ops-cpe-autoconfigure-00.txt

[Lorenzo Colitti <[e-mail]>]

On Mon, Jun 19, 2017 at 10:47 PM, STARK, BARBARA H <[e-mail]> wrote:

> The BBF flow has simultaneous RS and DHCPv6 solicitation. This is at odds
> with what you suggest. Simultaneous and sequential are both allowed by RFC
> 7084. I realize many cable providers have expressed a preference for
> sequential RS and DHCPv6 solicit, but many telcos have expressed a
> preference for simultaneous.


IIRC the preference for simultaneous was for telco environments where the
device had to emit a packet before the network would see it. The concern
was that because the device only sent three RSes on startup, if those three
RSes were dropped then it would never get an RA.

That concern is largely mitigated by resilient RS, which is now available
on fairly recent versions of Linux (and was backported to Android kernels
all the way back to 3.10, so CPE vendors can use it on older kernels too).


> The BBF flow mandates requesting IA_PD, while you propose it as a very
> weak MAY. Note that RFC 7084 mandates support for IA_PD. If you really want
> auto config of a CE router, this is the only way I know of to do it. I
> don't see that a MAY IA_PD requirement would ever result in auto
> configuration. Only a MUST will.


+1. An IPv6 CE router MUST do IA_PD otherwise it's not an IPv6 CE router.
It does not need to do DHCPv6 address assignment on the WAN interface
unless the ISP requires that for management purposes.


> BBF also supports auto-configuration via the unnumbered model where the RA
> does not include an Autonomous prefix and there is no offered IA_NA. RFC
> 7084 allows for the unnumbered model. You make no mention of the unnumbered
> model. I don't know how prevalent the unnumbered model is, though.
>

As a data point, I think most of the FTTH deployments here in Japan use the
unnumbered model.