The Network Map of the Internet
What does the internet look like? A few projects out there have approached the problem of creating a graphical representation of the internet - here are a few examples of the pictures you can come up with.
Far too often I see presentations from companies that show a computer in a home, connected to a modem, connected to this comic cloud shape labeled “internet”. Picturing the internet as an impending storm is eerily appropriate if you were a dot com in the 1990s, but the metaphor isn’t really very accurate.
In a recent class regarding the territoriality and legal regimes governing the internet, we were asked (as a conversation starter), “What does the Internet look like?” and “Where is the Internet?” So I looked into the question and found a couple of obvious answers, and here they are.
First of all, the Internet crosses borders effortlessly. This isn’t as facile a comment as it first seems. What I mean is that if the quickest (note: not shortest) way to route a communication from Miami to Seoul is via Cuba, that’s the way the routers will send it, and to hell with politics. This website is hosted in Livingston in the United Kingdom, but I am in Philadelphia, and most of my PHP code snippets are kept on a virtual drive somewhere else. This means that a map of the internet will be largely non-territorial in nature. Even if you can resolve IP addresses with country suffixes, just because a website ends in ‘.jp’ does not mean that the data you get when you look at it has come to you from Japan. The net result of all this is that it becomes extremely difficult to censor or control what is on the internet. Countries like China and Singapore try to do it by controlling every single access point that crosses their national borders. Whilst this is possible, it is both difficult and costly, because you have to funnel all the information through a restricted (controllable) number of channels. Even then, people can disguise packets of information going through those channels and sneak the information past the controls anyway.
The internet is a descendant of ARPAnet, which was built with redundancy in mind. That means that if you destroy a part of the network, the rest of the network routes around the damage to continue to function. It does this without the need to ask some central authority which way to go – the process is recursive and automatic; built into every router and every computer. The data’s most efficient path to its destination can fluctuate rapidly. Each document you send or receive over the internet is broken into “packets”, and these all flow through the system separately, possibly through different paths, and are reconstituted on the other end. So when you look at the map, you can draw different connections between any two points. The extent to which this is still true is debatable, with ultra-high bandwidth backbones carrying most of the traffic, but the network’s capability to do this remains, it’s just that when a backbone drops, the remaining routes in the network can no longer handle the high demand for traffic.
So what does a map of the internet look like? Well if you took every node connected to the internet and plotted how each and every single one was connected to all others, you’d need a very big piece of paper to plot it on. But if you set a program to plotting the nature of the internet that surrounds you, you can get something that looks a little like this (published in Wired Magazine in 1998)...
A stripped down map of the internet, showing just the links from my computer to a few others, looks like this...
But what of networks within networks? Take Gnutella, for example, a filesharing program that creates a peer-to-peer networking infrastructure. This uses the internet to connect, but doesn’t have the same fixed-point architecture; instead, in a way, it 'floats' above the internet. In other words, if you map the Gnutella network, two adjacent links on your Gnutella map could be at the opposite ends of the Internet map, but the network functions in a way that is very similar to the internet itself. These maps show personal connections forged between your computer and those of others, and how that network – which was dynamically created and constantly reforms itself as people leave and join – allows you to search the shared files of a computer on the other end of the network.
The GnuTella network’s dynamic and reforming structure makes it a very interesting phenomenon to study, especially since the introduction of “UltraPeers”, individuals who allow their node in the network to host a large number of “leaf nodes”, so that bandwidth needs can be funneled through volunteers who have bandwidth to spare.
A map of a small part of the GnuTella network as viewed from my PC is here…
So what does the entire internet look like? A few projects have drawn interesting pictures, and they sometimes look a little like this…
Links that may be of interest
Posted by nlvp at November 28, 2002 04:36 PM
