On Sun, 16 Oct 2005, Mike Leber wrote:
Does shim6 require new protocol stacks on the hosts at both ends of a session? (If not then the source is not making its own path selection decisions.)
As I understood it, shim6 is a way for two hosts to communicate between each other that they have multiple IPv6 addresses. So if a timeout occurs to the last used address, you can try another and try to resume the communication. So if the web-server has two different IP:s (from two different providers), both would be in DNS (preferrably) and the TCP session would be established with one of them. If shim6 detects that the original path is broken, it will try to use another and if it succeeds, the application won't notice anything as shim6 will abstract this to the TCP layer. I think this is a really good idea, having the network know about all multihomed companies just doesn't scale. With less prefixes and less AS numbers, network convergance would be much better. Think in the future, do we really want routers that'll handle millions of prefixes and hundreds of thousands of AS numbers, just because people want resiliance? If this can be solved on the end-user layer instead, it's more scalable. I can also see a loadbalancing scheme coming out on top of shim6 that'll be usable to end users as well. -- Mikael Abrahamsson email: swmike@swm.pp.se