Paul, I am talking about a "laptop" that connects an arbitrarily- complex internal network of virtual hosts and routers, and an arbitrarily-complex set of external devices attached on, e.g., Ethernet, Bluetooth, etc. I am attaching the diagram again (slightly updated) in case folks might have missed it in my earlier message.
So, it can't just be link-local-for-all, because then there is no opportunity for off-link communications when in fact the laptop may connect many links. Also, if my laptop ever needs to connect up with other sites (be it planned or ad-hoc; via phisical links or virtual) it will need to have something like ULA-C to avoid collisions. And, I don't want to have to inject a globally-routable prefix into the DFZ for it. Fred [EMAIL PROTECTED] > -----Original Message----- > From: Paul Vixie [mailto:[EMAIL PROTECTED] > Sent: Wednesday, June 27, 2007 1:14 PM > To: ipv6@ietf.org > Subject: Re: draft-ietf-ipv6-ula-central-02.txt > > > Discussions on this list seem to indicate that globally > routable PI might > > not be attainable for very small sites such as my laptop. > That would be an > > example of where I can't get my own PI prefix, right? > > > > [EMAIL PROTECTED] > > on a site small enough to be its own network (like your > laptop), i don't see a > use case for addresses other than ::1/128. ("host-local" is > a solved problem.) > > if you broaden your example to include multiple virtual > operating systems each > needing their own address so that they can communicate over a > virtual bridge, > then "lan-local" (fe80::/16) is available. > > if your laptop is joining an actual LAN (wired, wireless, > etc) then it will > have to have addresses assigned by that actual LAN's > administrator, which might > include both fe80::/16 and something else. > > it's in that final case where it's "something else" that the > question of PI > comes in. i don't think you're suggesting that your laptop > have its own PI > for self-communication, and i don't think you're suggesting > that your laptop's > PI ought to be connected by a routing protocol to the local network. > > at best your need for laptop-level PI would be so that you > could perform > routing over a VPN or tunnel whose endpoint was within the > local (actual LAN) > administrator's control. > > is this a use case worth pursuing for the purpose of defining internet > technology to support it? because it seems to me that the > mobile-IP folks > have scratched out a plan for this which involves using on your laptop > addresses assigned by the VPN hub, and speaking no routing protocol. > > -------------------------------------------------------------------- > IETF IPv6 working group mailing list > ipv6@ietf.org > Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 > -------------------------------------------------------------------- >
Egress Interfaces (to Internet) ^ ^ ^ | | | +------------------------+---+--------+----------+ | Internal hosts | | | | M | an routers | | .... | | A | ,-. | +---+---+--------+---+ | N | (H1 )---+ | | | E | | `-' | | +------+--< T | . | +---+ | | | | | . +--|R1 |---+-----+ | | I | . | +---+ | | Router +------+--< n | | ,-. | | | . | t | (H2 )---+ | Entity | . | e | `-' | . | | . | r | . | | . | f | ,-. . | +------+--< a | (Hn )---------+ | | c | `-' +---+---+--------+---+ | e | Ingress Interfaces | | .... | | s | (to internal networks) | | | | +------------------------+---+--------+----------+ | | | v v v Ingress Interfaces (to external networks) Figure 1: MANET Router
-------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------