On Tue, Jan 25, 2022, 5:00 AM Gorry Fairhurst <[email protected]> wrote:
> So... > > I've added something at the end, to note agree on the priority... > > > I'm really not sure I understand how these usecases are indended to work and will complicate an already complicated enough protocol. In particular discovering the indended proxy endpoint seems like a ball of wax. Then we have the issue of ossification as well as the privacy impacts of at least some of the proposals. > Other people have now mentioned enterprise networks; 5G (using a variety > of sub-IP technologies) and IoT, there are probably more. > I'm not really sure that these are the same problem. Advertising multiple subIP paths isn't the same as network based management, which really should be done at the host level. And ECMP in the network is a thing. It's possible to imagine opportunistic multipath that changes the 5 tuple hash to obtain multiple potential paths. In the case of satcom where FEC and retransmissions due to consistent packet loss together with bad latency are challenges for loss based congestion control the case for a proxy is a bit better assuming we can't succeed in having better congestion control algorithms. Sincerely, Watson
