Hi Brian, Hi guys, Brian E Carpenter <[EMAIL PROTECTED]> writes:
>> Water under the bridge and all that, but I'm curious: what was the >> rationale for this? It would seem to imply it is impossible to do >> incremental deployment of new headers, and thus essentially freezes >> routing header development for the lifetime of IPv6. >> >> Am I misunderstanding? > > I just looked this up in 2460 myself, and I'm jet lagged so I may > be confused, but indeed this seems to make it very hard to deploy any > new header except as a fork-lift upgrade among consenting systems. End-to-End. [with the deprecation of RH0,] IPv6 packets will flow from the source to the destination found in the IPv6 header, where the specific implementation will have the context and the code to analyse (filter) it and understand it (process the tiny details of the headers and upper layer). This leaves a simple work for the core: find the best path from the source to the destination, i.e. route the packet based on the content of IPv6 header destination address. End hosts then stack functionalities on top of that. The most low level (i.e. most demanding) functionalities start working in extension headers (MIPv6, ...) If someone comes with a _real_ need for having the core network providing help outside an efficient routing of its traffic to a specific destination, it will find help to make it happen but that's not the purpose of IP. Let's let people that still argue in favor of source-routing prove - there is a need - there is a need at L3 - there is a need at L3 to blindly process their pkts - there is a need at L3 to blindly process their pkts and it's safe And yes, deprecation removes blind source routing from IP stacks and leaves little room for having it coming back. Regards, a+ -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
