Telecommunications provider economics

Quite different models have applied to telephony versus Internet telecommunications provider economics. The models differ both at the individual user and the carrier levels. Traditional telephone calls, as well as conventional postal mail, operate on a "sender pays" model. Internet exchange, however, has been more mutual-benefit at the end user level, which can create an incorrect assumption that "the Internet is free".

Telephony sender pays & separations model

In the usual case, the caller will pay the retail charges for a telephone call. At the service provider level, however, very different models apply. Using a model called separations, telephone companies in a non-monopoly environment compensate both the callee's local telephone company, as well as intermediary long-distance and other providers, for carrying the call.

The example below uses some assumptions from U.S. telephony business models, but most countries that do not have a national telephone monopoly use similar techniques.

Caller's-----Caller's     Inter-      Inter-      Callee's
Local        Toll     ====mediate     mediate     Local
Service      Service      Provider....Provider    Service
Provider A   Provider     #1          #2      >>>>Provider B

In this example, the actual bill for the long-distance call comes from the caller's toll service provider, TSP. The customer also pays a monthly access charge to a local service provider, LSP-A, but the LSP-A charges are outside the scope of this example.

Assume that TSP charges $2.00 for the call. Either directly, or through various brokerages, IP1, IP2, and LSP-B expect to be compensated for carrying the call. Assume TSP keeps $1.00. TSP might pay $0.50 to LSP-B, and, in a direct payment model, pay $0.50 to IP1 and IP2. These payments are negotiated.

The separations model can produce some surprising effects. Many retirees live in the U.S. state of Florida, and many of those are parents who are dutifully called by adult children. There were occasions where the local access providers in Florida made more money from separations payments than they did from monthly access payments from their own customers.

It might be that TSP contracts with IP1 to deliver calls to LSP-B, and IP1 accepts the responsibility of delivering calls. In such a situation, there would be no direct payment from TSP to IP2; IP1 might pay, at a volume discount, $0.25 to IP2.

Given that there are hundreds or thousands of telephone companies in a large country, the negotiation and payments of settlements become extremely complex. As the Internet became more commercial, ISPs wanted to avoid this complexity.

ISP peering vs. transit model

ISPs related to one another in two basic ways, depending on whether they see the other as roughly equal in coverage, number of customers, etc., or if one connects to many more destinations than the other. While this was reasonably accurate in the early days of the Internet and its predecessors, things have become far more complicated.

Consider, for example, ISP 1, a small and home office provider, often of "triple play" Internet, television, and telephone service. If ISP 1 has a national presence, it can have very large numbers of customers who want to access content. ISP2, however, may specialize in serving content providers, such that it has a relatively small number of direct customers, whom ISP 1's customers want to reach. Are the two economic equals?

ISP as a customer

In contrast, a transit relationship involves the transfer of money from the customer AS to the provider AS; the transit provider accepts money for providing connectivity to all reachable Internet destinations. When an ISP, or an enterprise, "buys transit" from a provider. The provider will advertise its customer's routes to the rest of the Internet, as part of its service. In other words, the transit provider tells the Internet "I offer connectivity to this customer's piece of the address space." A transit provider will have connections to a large number of other providers, to which it advertises its customer space.

The terminology here is sometimes counterintuitive. Think of "advertising" as "attracting" traffic. "Accepting" routes from another AS, in social terms, is an invitation to a social engagement; the acceptor need not agree t that engagement, finding that a different advertisement is more attractive.

From the customer perspective, the simplest possible route acceptance relationship, exists when the customer has only a single "upstream" connection to the Internet, is to "accept default" from the transit provider. The default route is the route to which a router sends all traffic when it has no better match.

Whenever there is more than one physical connection to the Internet, the best load sharing and multihoming will take place when the customer receives full routing information from each connection. Having such information will allow the best exit to every destination to be chosen. The disadvantage of having this amount of routing information, however, is that the customer router has to have the memory and processing power needed to hold multiple views of the Internet, a different view from each point of connection.

Larger transit providers are often, but rarely exclusively, "facilities based", meaning they own and operate their own optical networks. In theory, if one bought transit from two major providers, there would be no common point of physical failure. In practice, providers often lease capacity from one another, so, unless the customer pays for facility diversity, there is no guarantee that two transit providers will have no common points of failure.

Peering assuming equality

If a pair of ISPs can agree that their customer base is approximately the same size, and it is of mutual benefits that their customers be able to reach one another. In such cases of peering,. When the parties agree that they are peers, they advertise, to each peer, the address space (i.e., of their internals and of their customers) to which they offer connectivity. No money changes hands because this is considered a balanced exchange.

To understand the nuances of these relationships, it is critical to understand that a peer will advertise to another, using BGP, only those destinations that belong to its customers or are part of its infrastructure. Even though a given may have connectivity to the rest of the Internet, it is not offering Internet-wide connectivity in a peering arrangement.