Hi, > It may be assumed that each upstream will be connected via a > separate router, not multihomed off the same router.
This seems to asume that one router equals one access network equals one provider. I don't think that's valid as a long term assumption. Any combination might arise - several upstreams via one router, different routers for different upstreams, and some upstreams via more than one router. In fact, > A host with mutliple upstream paths to the same destination (in- > home or external) should be able to use another in case on fails. seems to contradict the assumption. Or it's about multi-prefix based multihoming, which is a different problem altogether (see shim6 or draft-ietf-v6ops-ipv6-multihoming-without-ipv6nat). > Must support a "walled-garden" network. This might routing > based on either source address (from the walled garden network) > or destination address (to the walled garden network); support > for both is not required. I thought that the IETF was about the Internet, so we shouldn't be assisting the creation of walled gardens. However, source-address-based selection of the next hop is needed anyway, to support multihoming and deal with ingress filtering. I don't see how any scenario can avoid destination-address-based routing, so I think that support for both *is* required. Regards Brian Carpenter _______________________________________________ homenet mailing list [email protected] https://www.ietf.org/mailman/listinfo/homenet
