Fred, one by one, see inside. Heiner In einer eMail vom 19.05.2009 18:35:00 Westeuropäische Normalzeit schreibt [email protected]:
Heiner, > I have never understood either that inter- and intra-domain routing architectures have to be such orthogonal. > So whatever I proposed or supported so far was aiming to a consistent architectural framework, too. > Routing, which is not based on address summarization, would make sense inside of intra-domain-networks as well. Do these other approaches show how they recurse from the Internet core all the way to a singleton node as the limiting factor for recursion – or even to virtual networks within a singleton node? Abjuring the address asummarization paradigm in inter-domain routing, why should I stick to it in intra-domain routing ?:-( However there, the need for a fundamental change isn't as immense. Do the other approaches provide for discovery and utilization of multiple border routers? ... border routers as understood by each individual approach. But the answer is yes. Do they have fully-articulated specifications for automatic EID and RLOC address configuration? There is no need for such things. Do they support multihoming? What a question ! Of course, yes. Provider-independent addressing? PI- and PA- independent addressing, so to speak. IPv6 deployment? Yes, and whatever else. Traffic engineering? It would provide a large meadow for TE, never ever seen before. Secure redirection? promised. Ingress filtering? Given I know the topology and not only 300 000 routes, why should ingress filtering not be possible? Yes, it cannot be done as folks are used to. What about mobility? You'll bet. Without geographical coordinates, mobility wouldn't be. MTU handling for tunnels? Yes, I would also need some outer header, just like LISP. But there is also an alternative: Realization inside layer 2. This would shift the routing archtitecture from IETF to IEEE. Think about it ! Do the other approaches have their base mechanisms widely deployed in shipping implementations for many years, and with many millions of users?_ What if you don't need huge core routers anymore ? Is maturity important to this group? What about completeness? What about going new paths for this group ? Fred [email protected] ____________________________________ From: [email protected] [mailto:[email protected]] Sent: Tuesday, May 19, 2009 12:31 AM To: Templin, Fred L; [email protected] Subject: Re: [rrg] RANGER(S) In einer eMail vom 19.05.2009 01:26:57 Westeuropäische Normalzeit schreibt [email protected]: but in our experience RANGER(S) is the only proposal with a consistent architectural framework that applies recursively in a "network-of-networks" fashion from the global Internet core all the way outward to even the simplest of edge networks. I have never understood either that inter- and intra-domain routing architectures have to be such orthogonal. So whatever I proposed or supported so far was aiming to a consistent architectural framework, too. Routing, which is not based on address summarization, would make sense inside of intra-domain-networks as well. Heiner
_______________________________________________ rrg mailing list [email protected] http://www.irtf.org/mailman/listinfo/rrg
