Page 8 of 12
Internet pricing and interconnection
In traditional telecommunications, the cost of international connectivity has typically been shared, either by each party paying half the cost of the connection or by settlements based on the amount of traffic exchanged. Unlike the telecommunications industry, which has evolved complex systems of international charging agreements over a period of about 100 years, there has been no economic regulation of the Internet. The Internet industry is based on mutual peering and on an economic model of bilateral agreements between customers and providers. The result of this regime is that ISPs wanting to connect to the Internet backbone, usually in the USA, have to initiate the connection and often pay for the full cost of the circuit from their country to a large American ISP or Internet exchange. Many consider this situation to be unfair and a barrier to ICT development in many developing nations.
The following diagram shows a very simplified picture of the Internet industry and how traffic and payments flow when email is sent from one country to another. From the top down, the diagram illustrates how providers at each layer resell Internet connectivity to providers at the layer below. Connectivity flows down and money flows up. It is important to note that this model is independent of distance: end-users pay the same fee to access resources located on a server half the world away as they do for the webpage of a local newspaper.
The end-user buys connectivity from a local ISP. To carry the user’s email across the Internet, the local ISP sends it to an upstream provider, a national or perhaps regional provider that has a network connecting different towns and cities in that country or region. The local ISP, known as a tier-three provider, pays the larger tier-two provider for this connectivity.
To carry the email internationally, the national ISP routes traffic via global carriers, known as Internet backbones or tier-one providers. These are companies with high-capacity continental and international connections. Again, payment is made from the customer to the provider of service, that is, from tier two to tier one.
Tier-one providers connect with other tier-one providers, usually on the basis of peering. Their traffic flows are about equal, so it is mutually beneficial for them to simply exchange traffic as equals. Unless there is a large imbalance of traffic, tier-one providers do not usually pay each other. Instead, they operate on a “sender keeps all” model, keeping the fees they receive from the providers below them. After peering across the tier-one providers’ networks, traffic then flows downstream, from tier one to tier two and on to the end-user. But money only flows upstream. At each layer, the customer receives service from a provider and pays for that service.
Tier-one providers must make very significant invest-ments in network infrastructure, and the sender-keeps-all arrangement is the best way to recover these high costs. The result of this model is that developing nations and smaller ISPs must pay for the full cost of connectivity to the Internet, and they regard this as fundamentally unfair.
Comparison with the most commonly used traditional telecommunications settlement regime only makes matters worse. International telecommunications settlements tend to favour high-cost monopoly carriers over those operating at lower costs in competitive markets. Settlements are made on the basis of the volume of calls terminated by one country in another; and the payers under the regime tend to be developed nations, and the recipients developing nations. Settlements are made in US dollars and can amount to hundreds of millions of dollars each year. For many developing countries, telecommunications settlements are among their most important sources of hard currency.
The telecommunications settlement regime was already failing before the Internet emerged as a significant portion of international communications traffic. Commercial services such as callback and refile were developed to take advantage of the pricing imbalance created by the settlement regime. ITU estimated that callback accounted for 22 percent of traffic from Asia to the USA in 1995. So the suggestion that a settlement-type regime could be introduced for the Internet overlooks that the old telecommunications model was already failing before the Internet came along.
International charging arrangements for Internet services
The problem of Internet interconnection pricing is especially relevant to the Asia-Pacific region. It was first raised by the APEC Telecommunications and Information Working Group (APEC TEL) in 1998 in a study called International Charging Arrangements for Internet Services (ICAIS). The ICAIS study was based on the concern of Australian and Asian carriers that they are paying the entire cost of lines across the Pacific to reach the Internet backbone and resources in the USA.
ITU has now become the main forum where these issues are discussed. Unfortunately, most of the relevant documents are only available to ITU members (national governments and ITU sector members), meetings are typically for members only or for invited experts, and decisions are made by members. Three main types of connection relationship are being discussed:
The ITU working group is now trying to reach agreement between two proposed solutions. One is based on allowing market forces and negotiations between providers to determine appropriate interconnection rates and conditions (with a provision for development aid to support countries where there is market failure). This position is supported by “industry”, mainly large telecommunications operators. The second is a solution based on settlement peering where if a mutually satisfactory negotiated agreement cannot be reached the entities involved may use economic measures and traffic flow to determine who pays what. However, Internet traffic (packets) is much more difficult to measure than voice calls, and this seems to be the main sticking point in negotiations at the moment. This second solution has been supported by China and some other developing countries.
ITU established a study group to discuss international Internet connectivity in 1998 and has been unable to reach a solution. A new study period stretching from 2005 to 2008 was recently agreed. This is a critical issue, but one that is very difficult to follow as most of the discussions and documents are not publicly available. As an issue of importance to the Asia-Pacific region, encouraging trans¬parency in ITU processes would be beneficial.
Internet exchange points and regional backbones
The WSIS Declaration of Principles and Plan of Action avoid addressing the ICAIS issue directly and, instead, recommend measures to keep Internet traffic as local as possible as part of the answer to the problems of Internet charging and interconnection. They encourage the establishment of local and national Internet exchange points (IXPs) to keep traffic within the country that might otherwise be sent to the US backbone before returning, as well as the creation of regional Internet backbones so that traffic to neighbouring countries does not need to flow via more expensive international routes. It is important to note that when traffic is kept within a country the money paid for transport stays within that country.
IXPs can be established relatively easily and cheaply and can bring significant benefits to the local Internet in terms of cost, reliability, and ease and speed of connection. IXPs also aggregate demand for bandwidth and so are in a better position to negotiate rates for international connec¬tivity. In Asia Pacific, various network operator groups have been active and successful in deploying IXPs.11 There are important policy issues involved, particularly in countries where the communications regime is being liberalised and regulations on interconnection and the exchange of traffic may prevent service providers from using an IXP.12