Hi, Paul, On Sat, May 8, 2010 at 8:20 PM, Paul Jakma <[email protected]> wrote:
> Note that you can organise routing around AS numbers and so reduce routing > state requirements perhaps without having to change IP forwarding. I.e. you > can compute paths on an AS-based topology, and then compute IP routes based > on which ASes terminate which prefixes. As the association of IP prefix:AS > is a lot more static than the topology, you could reduce the amount of > information that had to be passed around. It seems your idea is based on global IP addressing. What I have in mind is to get rid of global addressing. So, only the pair (local) IP address + AS number would provide global uniqueness. IP addresses would be local to a specific AS. Same IP addresses can be reused in different AS's. This way we are free from the problem of address depletion. It doesn't matter whether you use IPv4 or IPv6 as your local addressing. A good thing, isn't it? So, IDR routers look only at AS numbers, not at IP addresses which bears no meaning in IDR. In fact, each AS(or AD) in DFZ can be seen as a node (point) in the IDR topology. So, the AS number can be interpreted as AS address (or AD address as Bob Hinden called it). Then, this topology is nothing different from those any intra-domain routing would deal with. Just another upward recursion of the same-quality routing domain. Therefore, you can even use OSPF or IS-IS in DFZ. I think this is what Heiner (or someone else) kept arguing for; why not use LS routing in DFZ? Remember Bob Hinden also commented on this possibility, too.. already back 15 years ago. How far sighted. And also, Bod Hinden commented that IS-IS provides an additional benefit of routing on endpoints, not interfaces or PoA. It is interesting that he even regarded it as benefit. We had our hero already since many years ago. And you know what? ILNP also says its ID points to the node, not to the interface or PoA. How nice. Now with bits from here and bits from there, the whole picture I had in mind seems to be converging to: o Addresses name the nodes(or end-points), not the interfaces or PoAs. o Node addresses are local to a given domain(or AS or AD...). o Domains are also named by domain addresses. Domain addresses bear context only in the inter-domain topology. o Each domain in the inter-domain topology reduces to a point or a node as called in usual routing topologies. Therefore, any routing protocol can be used, be it DV or LS. o Since domain addresses name the domain (point) itself, not the links connecting them, you might want to use IS-IS rather than OSPF since the former deals with endpoint addresses. In essence, we are reaching at a recursive network like Joe Touch or Jon Day asserted: o Within a domain, local addressing (for nodes) and LS routing (more specifically IS-IS since nodes are addressed, not the interfaces) are used. o At a higher recursion level formerly known as DFZ, also local addressing (for domains) and IS-IS are used. o Repetition of exactly the same routing infra in two adjacent layers. o What if we have to deal with inter-planetary routing? Or when we face with the address depletion problem also with Domain addressing? Then, build another layer ontop, and repeat the same routing infra. Out of the solar system....? o What if we have to dig into our human body to extend our networking internally into our organs and even cells?.. body area network... Then you just repeat the same routing infra in the reverse direction or in the inward direction. The one point node now becomes a universe for a population of internal sub-node points. o Just like the same nature repeats inwards to micro-cosom and outwards to ever expanding universe. Isn't this simple? Isn't this scalable? We don't need to manage global unique address spaces. We don't need governance. We are now free. Our systems is fully decentralized, distributed. > > There's still a lot of details and problems you'd have to work out though. > > regards, > -- > Paul Jakma [email protected] Key ID: 64A2FF6A > Fortune: > "The Computer made me do it." > -- Regards, DY _______________________________________________ rrg mailing list [email protected] http://www.irtf.org/mailman/listinfo/rrg
