20.6.07

La topología de Internet

> La revista Cosmos publica una descripción de Internet, de cómo se parece físicamente, su topología. Resulta que es una gran red en forma de esfera con un centro sólido.

1 comentario:

Andrés Hax dijo...

News
Internet is spherical with a dense core
Tuesday, 19 June 2007by Hamish Clarke
Cosmos Online

Visual representation of the topology of the internet. Data was collected by thousands of world-wide volunteers through a distributed computing project called DIMES. Color code indicates a relative measure of node importance, with nodes in the most inner shell (red) forming the nucleus of the Internet.
Image: Yuval Shavitt (Image was generated with the Lanet-vi program of I. Alvarez-Hamelin et. al.)

advertisementSYDNEY: The internet is not web-shaped after all. Instead, it's more more like a globe, made up of a dense core and sparsely connected outer regions, according to a mathematical analysis of its structure.

Analysing the shape, or 'topology' of the internet may seem like an esoteric activity, but according to software engineers led by Yuval Shavitt, from Tel-Aviv University in Israel, it has practical implications.

"A better understanding of the internet's structure is important in many ways," said Shavitt. "We can track the internet's evolution in time and generate predictive models to its growth [which will] help us understand how the Internet will look a few years from now so we can plan ahead."

The findings are published this week in the U.S. journal Proceedings of the National Academy of Sciences.

Rapidly evolving structure

Scientists from a range of disciplines are increasingly seeking to understand the internet's rapidly evolving structure using mathematical tools with exotic-sounding names like scaling theory, percolation and fractal analysis.

For their analysis of how the internet's 'nodes' (internet service providers and large organisations with a lot of web traffic) are connected, Shavitt's team borrowed a method from mathematical graph theory known as k-shell decomposition. Using this method, each internet node was placed in a group, according to the number of links it had to other nodes.

The researchers discovered that the Internet consists of a dense, heavily connected nucleus of about 100 nodes, including Google and U.S. telecommunications giant ATT WorldNet. Surrounding this core was a region they called the 'peer-connected' component, which is able to connect to the bulk of the Internet without causing congestion in the nucleus. It is possible for data to get between any two points in the peer-connected component within about four links. The final, outermost layer is more sparsely connected, and must travel through the nucleus to reach other nodes.

The authors recommend that in the future, communications should be more often routed through nodes in the peer-connected component to improve the Internet's overall capacity and to avoid congestion in the nucleus.

According to Shavitt, a better understanding of the Internet's topology may lead to improvements in applications that are already in use today, "such as Web surfing, file sharing, VoIP [voice-over-internet protocol, such as Skype], and video/voice streaming applications."

Money, money, money

Not everyone is in agreement though.

"I believe the story should go the other way: the topology is driven by the application need, not the other way around," said Lixia Zhang, a computer scientist from the University of California in Los Angeles, U.S.

Zhang believes that Shavitt and his team may have gotten things backwards in this respect. However, she believes a more fundamental flaw besets this research.

"The study seems to treat all connections between nodes as the same, and focuses mainly on the node degree [of connectivity]," Zhang said. "This is just plain wrong." Zhang argues that Shavitt's study overlooks important distinctions between so-called "customer" and "provider" nodes.

One thing Zhang and Shavitt agree on is that money makes the World Wide Web go round. Both say that the biggest driver of the Internet's rapid evolution now is the interplay between technology and the economy, as more and more new companies try to become the next Google, eBay or Facebook.