On Oct 3, 2011, at 9:00 PM, Thomas Herbst wrote: > > There will be wide area network providers who interwork with the home network > but do not provide global connectivity. > Two mentioned so far are utility networks and 3g providers. One of the > outputs of the wg should be to define how they should be configured to > perform their role without messing up Internet communication.
Those utility networks have a fundamental problem that I contend is beyond the scope and charter of HOMENET. Utility networks of that sort do not provide transit to the Internet default-free zone. They must therefore obtain their routes to residential networks bilaterally. This implies that these utility networks could be-- and would do well to be-- numbered with ULA prefixes, and that they should use of an exterior gateway protocol at their border with residential networks so that each home network can advertise, into the utility network, its list of globally assigned prefixes that it obtains from Real Internet Service Providers. Yes, the scale of the routing problem faced by these utility networks is hard. No, I don't think they're going to be able to solve it adequately. This is-- however-- not our problem. It's theirs. They would not have this problem if they engineered their networks differently, i.e. to rely on real Internet service providers to provide transit through the default-free zone between their equipment in residential deployments and their equipment in their data centers. This is what EVERYONE ELSE in the world does, and it works pretty well. > A wg Chair from the Internet area did accuse me of "breaking the Internet > model" because the utility networks my company builds do not provide global > connectivity to users with our 100kb to the node. That's really not the problem, if you want my humble opinion. The problem, I would say, is that these utility networks insist on extending their private routing domains into our home networks where they don't belong and they aren't welcome. On Oct 3, 2011, at 6:46 PM, Erik Nordmark wrote: > [...] > 6. The lookup of foo.ispA.net works over either DNS and returns the same IP > address, but the application-layer content is completely different (e.g., a > "subscriber" view when connecting over the ISP-A connection). This is the basic problem faced by any multi-homed host, e.g. a personal computer with a 3G interface and a Wi-fi interface that are simultaneously active, along with a split-tunnel VPN interface [or three] running on one or both interfaces. It is a problem for host operating systems and applications developers. I suggest HOMENET should steer well clear of it, and just about every related problem that is too easily conflated with it. -- james woodyatt <[email protected]> member of technical staff, core os networking _______________________________________________ homenet mailing list [email protected] https://www.ietf.org/mailman/listinfo/homenet
