A previous post, on “Barabási – Albert preferential attachment and the Internet“, gave a plot of the Internet, as a sparsity map of its regular adjacency matrix, with the axes ordered by each ASes eigencentrality:
Sparsity plot of the Internet adjacency matrix, for the UCLA IRL 2013-06 data-set, with the nodes ordered by their Eigencentrality ranking.
Each connection in the BGP AS graph is represented as a dot, connecting the AS on the one axis to the AS on the other. As the BGP AS graph is undirected, the plot ends up symmetric. The top-right corner of this plot shows that the most highly-ranked ASes are very densely interconnected. The distinct outline probably is indicative (characteristic?) of a tree-like hierarchy in the data.
Who are these top-ranked ASes though? Are they large, well-known telecommunications companies? The answer might be surprising.
The following table lists the highest eigencentrality-ranked ASes in the 2013-06 BGP AS graph from the UCL IRL Topology data-set, along with their ASN and degree (out of 41267 non-0-degree ASes):
I can recognise at most a few of these networks. The rest I do not, and I assume are more regional players. I certainly wouldn’t have expected a Swiss schools network to be in this list. Interestingly, a majority, 11, of the top-20 Eigencentrality ranked networks in the above table appear clearly to be Europe based, and another 6 in western Russia, whose networks are likely to gravitate toward peering in western Europe (the European RIR, RIPE, is the RIR for Russia).
What then of the major telecommunications companies? Where do they rank? This is shown in the following table, listing some well-known ones, as determined by examining the Wikipedia page on Tier-1 ISPs and the annual Renesys report on the “Baker’s Dozen”:
A few of these major telecommunications have their operations split over multiple ASNs, notably Level-3, and might rank more highly otherwise. However, generally they all rank quite lowly, despite many having degrees at least as high as those top eigencentrality-ranked nodes.
So what’s going on? The answer, likely, is the IXP model prevalent in Europe, where smaller networks meet and peer very widely. This allows them to avoid paying the traditional telecommunications companies to exchange traffic. The larger IXPs reportedly carry as much daily traffic between the networks peering at them, as the largest ISPs do over their backbones. These IXPs also enable major content providers to peer directly with a great many networks, and there is evidence these major content providers have built out their connectivity widely in this way, rather than relying on the traditional large telecommunications companies to route their traffic for them.
IXPs and the large, densely peered meshes of smaller networks they enable seem now to be at the heart of the Internet. The structure of the Internet appears to have evolved, as well as our understanding of it, away from a tiered hierarchy of ISPs.
Paul Jakma's Blog 
Leave a Reply