Internet architecture is organized into several planes. The data plane represents how data is actually forwarded between routers. Topologies based on the data plane reflect physical nodes and connections but these topology graphs are hard to create and harder to validate for correctness or completeness. The control plane represents how ISP's create and configure organizational routing policy. While the data plane is responsible for forwarding data based on policy, the control plane is responsible for configuring policy. As far as topology is concerned, nodes in this plane are autonomous systems, (AS) and links between nodes represent some type of relationship. There are two types of relationships. Customer-Provider links involve one AS providing Internet access to a smaller AS at a monetary cost. Peering links are agreements between two AS networks to exchange certain types of traffic free of charge. These types of Internet topologies have become a hot area of research in academia. There is a third less known plane called the management plane but the topology maps it produces are incredibly inaccurate since its data can be directly modified by network operators with no effects on routing dynamics.
Fortunately, nobody owns the Internet, there is no centralized control, and nobody can turn it off. Its evolution depends on rough consensus about technical proposals, and on running code. Engineering feed-back from real implementations is more important than any architectural principles.— B.Carpenter, Architectural Principles of the Internet; June, 1996.
The simplistic Jellyfish model of the Internet centers around a large strongly connected core of high-degree web pages that form a clique; pages such that there is a path from any page within the core to any other page. In other words, starting from any node within the core, it is possible to visit any other node in the core just by clicking hyperlinks. From there, a distinction is made between pages of single degree and those of higher order degree. Pages with many links form rings around the center, with all such pages that are a single link away from the core making up the first ring, all such pages that are two links away from the core making up the second ring, and so on. Then from each ring, pages of single degree are depicted as hanging downward, with a page linked by the core hanging from the center, for example. In this manner, the rings form a sort of dome away from the center that is reminiscent of a jellyfish, with the hanging nodes making up the creature's tentacles. More info on Jellyfish Concept and Image can be found on the following Web Page: http://www.mundi.net/maps/maps_020/
Bow Tie Model
The Bow Tie model comprises four main groups of web pages. Like with the Jellyfish model there is a strongly connected core. There are then two other large groups, roughly of equal size. One consists of all pages that link to the strongly connected core, but which have no links from the core back out to them. This is the "Origination" or "In" group, as it contains links that lead into the core when originating from it. The counterpart to this is the group of all pages that the strongly connected core links to, but which have no links back into the core. This is the "Termination" or "Out" group, as it contains links that lead out of the core and terminate from it. The fourth and final group is all other disconnected pages, which neither link to the core nor are linked from it.
A more comprehensive representation of the Bow Tie model has been presented with additional, smaller groups of web pages. In this version, both the "In" and "Out" groups have smaller "tendrils" leading to and from them. These consist of pages that link to and from the "In" and "Out" group but are not part of either to begin with, in essence the "Origination" and "Termination" groups of the larger "In" and "Out". This can be carried on ad nauseam, adding tendrils to the tendrils, and so on. Additionally, this more detailed version contains another important group known as the "Tubes". This group consists of pages accessible from "In" and which link to "Out", but which are not part of the large core. Visually, they form alternative routes from "In" to "Out", like tubes bending around the central strongly connected component.
- Siganos, Georgos; Sudhir L Tauro; Michalis Faloutsos (Dec 7, 2004). "Jellyfish: A Conceptual Model for the AS Internet Topology" (PDF). Retrieved 2007-12-29.<templatestyles src="Module:Citation/CS1/styles.css"></templatestyles>
- "IBM Almaden - News - Researchers Map the Web". Retrieved 2008-11-11.<templatestyles src="Module:Citation/CS1/styles.css"></templatestyles>
- "Link Popularity and the Bowtie Theory". Retrieved 2008-11-11.<templatestyles src="Module:Citation/CS1/styles.css"></templatestyles>
- Internet Topology Mapping Tools (Mehmet Engin Tozal)
- The Workshop on Internet Topology (WIT) Report
- Perl program that generates synthetic Internet-like topologies
- Computing the unmeasured: An algebraic approach to Internet mapping