Question: Who pays for the Internet? Several readers, including Max Hahto <[email protected]> and Yves Dussault <[email protected]>, have written to ask how the Internet is paid for. That is, since the Internet is worldwide, there must be long distance or other charges involved in moving data from your dial-up or permanent connections to other locations. How does this work?
Answer: The Internet is built a bit differently from the phone company’s voice network. Voice calls are created as a "circuit" that is dedicated through a series of phone company switches from the caller to the receiver. This circuit remains open and in use – and reserves some of the network’s capacity – for as long as the call lasts.
The Internet, by contrast, is packet based. We’ve talked about packets in previous issues (see Hey, I’m Talking to You! in NetBITS-001 and NetBITS-002); they’re a way to split up streams of data into bite-sized units, usually a few hundred bytes long, that can go across different paths but still arrive at the same destination.
Because of this, your local dial-up call that goes to an Internet service provider (ISP) connects over a voice-type circuit. Once it reaches the modem on the other end at the ISP’s point of presence (POP), it’s transferred to the Internet. The ISP pays a combination of fixed and metered charges to aggregate data from their POPs to other locations over high-speed lines that can handle the equivalent of dozens or even thousands of voice calls. They pay charges based on distance and capacity, but those charges are typically fixed.
Sometimes ISPs must pay for data transmission over certain amounts if they buy a lower-tiered service but want to keep extra capacity on hand in case they experience a growth spurt or a busy day.
Other Internet providers who offer services to businesses – we call these network service providers (NSPs) or mid-level networks – as well as ISPs all connect their traffic to exchange packets bound for other networks at several points around the country. Some of these are Network Access Points (NAPs), which are lightly funded by the National Science Foundation. These NAPs and similar "peering" points are run by telephone companies (telcos), such as Ameritech, Worldcom, and Pacific Bell.
The "peers" who want to exchange data in these points pay fees to the telcos for hosting their equipment in these physical locations and running most of the infrastructure for them. Anyone can run a line into a peering point, but they must negotiate agreements with the other companies in those locations to exchange traffic with them.
The Internet is made up of these kinds of deals and peering points: it’s a collection of different networks that agree to treat each other as though they were part of a cohesive whole. So any network, down to an individual who’s dialed up at 14.4 Kbps using PPP, up to IBM’s worldwide WAN (Wide Area Network), is part of the Internet. The Internet doesn’t exist as a separate entity from the machines and routers that collectively make it.
Thus, the money comes in at every point at which data is passed up and down, and most of it is collected by ISPs, NSPs, and telcos. The telcos make their money from extra phone lines, the high-speed dedicated digital lines, and the peering point. ISPs and NSPs – and large companies with their own nationwide or global networks – pay these fees to telcos, but they collect fees from their downstream customers.
Ultimately, as in all business, the downstream fees they collect from their users and customers should more than pay for the upstream fees they pay for transiting their information as part of the Internet.