Re: a draft about on-link and submit prefixes

[james woodyatt <[e-mail]>]

Hi everyone,

First, my thanks to Jinmei-san for this well-written draft. I have only minor quibbles with it, most of which were already surfaced by Lorenzo, and the rest are probably nits not worth mentioning at this point.

I’d like to respond to a comment that came in the ensuing discussion. I think it’s relevant.

On Mar 14, 2017, at 06:20, [e-mail] wrote:
> 
> I haven't seen a definition of what an IPv6 subnet is. I specifically
> haven't seen where it is defined that a given subnet prefix can be
> spread across multiple links - especially since RFC 4291 says
> 
> "Currently, IPv6 continues the IPv4 model in that a subnet prefix is
> associated with one link."
> 
> So if this is no longer the case (just like it's not the case that all
> IPv6 netmasks are /64 or /127) then it is useful to have this defined
> somewhere.

p1. In my proposal for additional text in revising RFC 4291, I noted that the sentence highlighted above is a) superseded by The IPv6 Subnet Model [RFC5942], which coincidentally provides a definition of an IPv6 subnet in section 1, and b) I offered an alternative formulation of that text with the statement updated to match current reality accordingly.

p2. A fine example of an existing technology where a subnet is shared across multiple links is "Extending an IPv6 /64 Prefix from a Third Generation Partnership Project (3GPP)  Mobile Interface to a LAN Link” [RFC7278]. Another that comes to mind is a certain application of "Compression Format for IPv6 Datagrams over IEEE 802.15.4-Based Networks” [RFC6282] at HOMENET routers connected to residential LAN networks behind IPv6 Customer Edge Routers that implement "Basic Requirements for IPv6 Customer Edge Routers” [RFC7084].

Clearly, that sentence in RFC 4291 highlighted above, which persists in the latest draft of I-D.ietf-6man-rfc4291bis too, is a source of confusion. We have published many RFCs, several of them in the Standards Track, that conflict absolutely with that statement. It’s definitely not an accurate statement anymore, if it ever was.


--james woodyatt <[e-mail] <mailto:[e-mail]>>