The Internet backbone is composed of high-bandwidth lines interconnected with fast, high-capacity routers.
Backbones can span large geographic areas. The backbone lines, generally installed by national long-distance telephone carriers or by the government, employ the highest speed transmission paths.
Network access points, or NAPs, tie together various ISPs and the national telecommunications companies.
The Physical Structure of the Internet
The Internet has a loose structure. Its topology is mainly a consequence of perceived commercial need being satisfied by a network service.
The result is a broadly hierarchical structure, however, with some operators concentrating on providing core networks, others concentrating on connecting private users, and still others providing transit between users and the core.
Figure 3-2 illustrates a possible configuration.
The core networks (called national service providers [NSP] in the United States) compete with each other to carry Internet traffic.
They include internetMCI, Sprint- Link, PSINet, and UUNet Technologies.
They have high capacity links between their routers, capable of operating at gigabits per second and more. NSPs connect with each other in two ways:
Through independent "network access points" (NAP). These are commercial operations and provide high bandwidth switching between networks.
They are generally confined to a small geographical area (i.e. single building) and consist of high-speed networks such as gigabit Ethernet that host routers belonging to the NSPs. NAPs are sometimes called metropolitan area exchanges (MAEs).
Through private "peering" arrangements. These are direct connections between routers belonging to the NSPs.