FWIW, based on my skimming of the document, I think this is a bad approach. Reasons:
1) the scenario is rather limited, a single link directly connected to multiple ISPs (through different routers). IMO, this is too limiting a scenario to develop a specifiic solution that doesn't address a other scenarios. 2) It places additional complexity in the hosts, and it is not clear to me that the benefits are sufficient. Indeed, it's not immediately clear what the "multihoming" aspect is. In my mind, "multihoming" means having multiple ISP connections, and actually using them simultaneously. The approach in the document seems to consist of "pick just one", which is not multihoming IMO. Or if it is "use multiple ones simultaneously", then the host needs to be sure to route packets to the right router based on the source adddresses used. That is signifificant (new) complexity and I am not at all sure we should be doing that at this point in time. So, at this time, color me very skeptical to this idea. Indeed, I'd prefer backing up a bit and having discussions about the problem statement and what the actual problem is that needs solving. Thomas -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
