Re: 64share v2

[[e-mail]]

Hi Lorenzo,

>  - the PIO is used for prefix discovery on the directly connected link, a different option
>    "delegated prefix option" should be created. A principle of good protocol design is to be explicit
> 
> +1. We can't just change the length of the PIO from 64 to 60, since this is really not the same as a PIO. A PIO specifies "the prefixes that are on-link and/or are used for stateless address autoconfiguration". Here, the prefixes are not on-link, and any non-/64 PIO can't be used for autoconf anyway, because for global addresses, IIDs are 64 bits long. If we just change the PIO from /64 to /60, IPv6 on existing mobile devices will just stop working. Also, the A and L bits in the PIO are meaningless. If the network is delegating a /60 prefix to a mobile device, it makes no sense to say "you can have this entire /60 but nothing in it can use autoconf", or even "you can have this /60 and autoconf is available on it".
>  
>  - ND has been designed to be datalink type agnostic. This mechanism should not assume a 3GPP link.
>    Needs text to state that this new option requires per client messaging or an underlaying p2p link.
> 
> +1
>  
>  - Unsure if you need some way for the client to refresh or request a delegation.
> 
> Personally I don't think we need a way to do this. For the problem at hand, the semantics of RAs are fine, it's the semantics of the PIO that don't meet the needs.
>  
>  - You need to clarify host/router behaviour in that you now make the RS/RA behaviour between routers.
> 
> Not sure that necessarily needs to be added here. After all, we published RFC 7278 without it, and the problem is basically the same. And RFC 6204/7084 (IPv6 CE routers) already expects such routers to listen to RAs.
>  
>  - You should require a mechanism for the CE to verify forwarding state in the PE. We missed that in 3315.
> 
> I don't think this is a problem in mobile networks. Certainly, I don't think we currently have anything along these lines for IPv4 addresses or IPv6 /64s.

Right, you do get a very clear L2 event on mobile networks.
If we want to make this general it might be needed in other networks.
I thought it might be something to consider, given how many problems we've seen in broadband deployments, where the PE does DHCPv6 PD snooping as a relay, and seems to forget state. And unless the CE actively probes there's no way to recover.

> Instead, I think we should say that the prefix MUST be a superset of a prefix that already has a PIO with A=1. That would make failure modes very clear.

Implied exclude option. Yeah.

Cheers,
Ole