A Network of Networks

As I described in Chapter 2, a local area network is a group of computers connected together so that data can move among them. In the simplest terms possible, the Internet is a network of networks rather than a network of computers. It's a general approach to moving data among separate local area networks, some of which may be half a world apart.

Odd as it may seem, the Internet grew out of a desire to create a communications system that could survive nuclear attack. Not surprisingly, the Pentagon was the driving force in the early years, accompanied by the RAND Corporation and researchers at universities and large technology companies like Xerox Corporation. A military group called the Advanced Research Projects Agency (ARPA) worked with the civilian sector in the late 1960s to design and test ways of passing data among widely separated nodes. The data had to be sent in such a way that if one or more nodes were destroyed by military action, data could be routed around the damage and still be delivered to its destination.

What was originally called ARPANET first went live in 1969, and grew quickly and steadily throughout the 1970s, even though it was still limited to government, university, and corporate research organizations. In those years there were no personal computers, and 'local area networks' were small groups of large and extremely expensive mainframe computers and minicomputers, which were 'mini' only in comparison to the mainframes. (Minicomputers were 'only' the size of refrigerators rather than the size of New York studio apartments.) Being on ARPANET was a fairly rare privilege back then, and remained so until ARPANET 'went public' and became the Internet in 1989. Even then, the general public had a hard time finding a way onto the Internet until 1992 or 1993, when Internet access exploded into a growth business.

The notion of the Internet as a network of networks is subtle. From where you sit at your computer, it looks like the Internet is a gigantic local area network of countless separate computers. The Napster file-trading program was used by tens of millions of people, all connecting directly from one computer to another to trade MP3 music files. Nonetheless, the Internet is really a way to connect networks -and then it's up to the networks to connect to individual computers.

Backbones and Interstates

A simple metaphor for the Internet is the Interstate Highway System. All across the United States are cities and towns of many sizes. Each city is a cluster of homes and businesses, arranged on a network of streets and roads. Connecting many of these cities are the huge highways we call the Interstates. There are no homes nor businesses built right on the Interstate highways. The highways are strictly for moving between cities.

So it is with the Internet. There are a fair number of high-volume data conduits called backbones running around the United States and much of the rest of the world, connecting networks into the Internet. In very simplified form, the structure of the Internet is shown in Figure 3.1.

Figure 3.1: Interstate Highways for Data.

Each of the bubbles in Figure 3.1 represents a network, and the lines running between them represent the various Internet backbones. Each of the networks is a group of computers belonging to one organization or individual. They can be of various sizes, from immense (imagine how many computers Microsoft or IBM has) down to a 'network' of only one computer.

The networks are very much as I described them in Chapter 2, except that each has something new this time: a router. A router is a device that facilitates communication among networks. It accepts data from one of the computers on its own network, and routes that data to the network containing that data's destination computer. A router's job is to know where other networks are and how to get data to the next router on a trip that may involve numerous hops among several routers.

Not all networks are connected directly to all other networks. A path from one network to another may require intermediate stops at a fair number of other networks. In Figure 3.1, getting from Network 1 to Network 3 requires an intermediate stop at either Network 2 or Network 4. Although one of those two paths may be by convention the 'normal' path, either path will do. And if someone drops a nuclear bomb on Network 4, Network 2 can take over and get the data where it needs to go. It's possible to blow holes in the Internet, and whereas the Internet might slow down as some of its capacity is destroyed, it would be very hard to bring it completely to a halt. The Internet is very 'fault tolerant,' and although it has weaknesses (like 'denial of service' attacks-deliberately caused packet floods) dependence on a particular set of physical connections is not one of them. Any connection will do, and the Internet is an ace at finding connections.

Networks within Networks

Routers rule the Internet. The Internet is, in fact, a kind of 'router's club,' in that the routers are the ones that know how to get data from here to there, and cooperate tirelessly in making all necessary connections happen. In fact, routers are so good at managing traffic between networks that they are also used to manage traffic within networks, by cutting up large networks into smaller subnetworks, with a router governing communications at the gateway to each subnetwork.

The best example would be a mid-sized company that is divided into departments. The whole company might have 200 computers, but there are four departments, each with 50 computers. Most of the traffic that moves from computer to computer stays within department boundaries, simply because people who do similar work tend to work on similar data-and often the same data. Each department has a router on its network, and that router confines departmental network traffic within the department, unless it explicitly has to go outside the department. Instead of one huge network of 200 computers, there are now four smaller networks of 50 computers, acting mostly independently.

The Internet began as a peer network, in which all networks stood as peers to one another. As the number of both networks and computers attached to networks grew stratospherically, the Internet became a hierarchical network, with immense networks containing large networks containing small networks. Big routers handle communications among networks, and smaller routers handle communications between the smaller networks contained by the big networks. So it goes, with networks inside of networks, smaller networks containing still smaller networks, until at some point we find you.