The Internet is an international “network
of networks.” In order to provide the physical connections between widely separated
broadband resources and consumers, countries must establish international links
(gateways) to connect to the world’s Internet and telephone networks. The technologies
providing long-haul transmission, such as fiber optic cable and satellites, typically
have very high investment costs. While initial “sunk” costs are high, they have
very low incremental costs to accommodate additional users. These technologies also
enable carriers to activate additional capacity on an incremental, graduated basis
as demand grows.
5.4.1 International Links
The vast majority of international telecommunications
traffic is carried by undersea cable systems—more than 95 percent according to some
estimates (Bressie 2010). This reflects the advantages of fiber optic cable in terms
of bandwidth and latency compared to satellite. Undersea fiber optic cables can
transmit data at speeds measured in Tbit/s, while even the newest communication
satellites offer speeds below 1 Gbit/s as well as higher latency. As of early 2011,
there were more than 120 major submarine cable systems, with another 25 planned
to enter service by 2015.9
Submarine cables are quite expensive to deploy, with costs that routinely reach
into hundreds of millions of U.S. dollars. As such, many are financed by consortiums
of operators rather than a single investor. For example, the Eastern Africa Submarine
Cable System (EASSy) has landing points in nine countries and connects to several
additional landlocked countries; it is funded by 16 African and international shareholders,
all of whom are telecommunications operators and service providers.10
While undersea fiber
optic cables may be the preferred option for international connectivity, it is not
a viable option for some countries and operators. Landlocked countries, for example,
do not have direct access to the sea and thus are constrained in their ability to
exploit submarine technology fully. Transit costs to tap into an undersea cable
can be significant (national and regional fiber backbones may not be available to
tap into the undersea cable), but this is becoming less of an issue over time, as
landlocked countries complete some type of fiber connection to international cables
through neighboring countries. Landlocked countries may be able to negotiate a virtual
coastline so that they own and operate a cable landing station in a neighboring
country’s territory but otherwise depend on the neighboring country to provide reliable
and reasonable prices for transit. Many small island developing states (SIDSs),
mainly in the Pacific Ocean, are distant from undersea fiber routes, and the economics
of connecting to undersea cable are problematic. Regulatory restrictions or high
costs may restrict service providers from accessing undersea cables. These factors
tend to encourage the use of satellite connectivity. Another issue is that, even
where countries have access to undersea cable, they still may want to deploy satellite
as a backup to ensure redundancy.
Service providers need
to contract physical international connections in order to support their end user
broadband requirements. They do so either by participating in ownership consortiums
of the physical facilities or by leasing connectivity through wholesale operators.
A relatively small number of Internet service providers (ISPs)11 have the financial resources
needed to invest directly in capacity in international backbone broadband networks,
so most lease capacity from larger international operators. This can present several
business and regulatory challenges:
Monopoly or dominant control of international backbone routes. Physical backbone networks
are generally owned by a few operators or consortiums. The restricted ownership
can be a barrier for nonaffiliated ISPs that need international connectivity.
Monopoly or dominant control of international landing points. Landing stations for
undersea cable and satellite earth stations are generally controlled by a few entities.
Even if an ISP has successfully contracted for capacity on a fiber optic cable or
satellite transponder, it may be constrained in its ability to connect that capacity
to its domestic backbone network. For example, a service provider may lease capacity
on a satellite transponder, but may have to come to a separate agreement with the
owner or operator of the international gateway that receives the satellite’s transmission.
The potential for international connectivity
to be a bottleneck in the development of broadband connectivity cannot be overstated.
Submarine cables connect to domestic backhaul networks at a cable termination station,
which is—but may not be—the same facility as the cable landing station (that is,
where the cable makes landfall). Because all operators in a market, particularly
new entrants, may not have the resources to own and operate a cable landing station,
the owners of such stations—generally the incumbent operators in newly liberalized
markets—may be required to provide access to the station, and therefore to the cable,
on reasonable terms to competing service providers. Limited access to landing stations
can have a chilling effect on the diffusion and take-up of broadband services. Conversely,
limited opportunities or burdensome regulations related to cable landing can discourage
interest in that market among cable operators, again creating a connectivity bottleneck.
Governments and regulators may need to implement competitive policies with respect
to issues such as submarine cable landing stations, open access, and infrastructure
sharing to eliminate such bottlenecks (see chapter 3).
In Singapore, for example,
the single cable landing station was owned by the incumbent operator at the time
of market liberalization. Singapore’s regulator undertook two parallel approaches
to improve international connectivity (IDA 2008). The Info-communications Development
Authority of Singapore (IDA) required the incumbent operator to offer collocation
in its submarine cable landing station to alternative operators, later imposing
connection at regulated prices and granting alternative operators access to the
capacity of submarine cables on behalf of a third party. In addition, the IDA streamlined
the administrative procedures for submarine cable companies to obtain landing permits
and authorizations in Singapore. As a result of these and related actions, prices
of international leased circuits in Singapore decreased 95 percent, total submarine
cable bandwidth capacity increased from 53 Gbit/s in 1999 to 28,000 Gbit/s in 2007,
and broadband penetration reached 77 percent of households in 2007.
5.4.2 Internet Links
Whether via fiber optic cable or satellite,
securing physical international links is only the first step in procuring international
Internet connectivity. ISPs also need to arrange for exchanging and routing their
traffic. Such arrangements ensure that Internet traffic can be delivered anywhere
in the world, eliminating the need to have physical connections to every country.
An ISP will typically arrange to hand off its traffic at the points where its contracted
physical connectivity terminates. Such arrangements are usually of two kinds:
A peering arrangement is where two ISPs freely exchange Internet traffic.
The peering requirements of large ISPs often exceed the capability of smaller ISPs.
For example, in order for an ISP operating in the Asia-Pacific region to peer with
Sweden’s TeliaSonera, it would have to provide traffic equaling at least 500 Mbit/s
and the ratio of inbound and outbound traffic exchanged between the ISP and TeliaSonera
could not exceed 3:1 (TeliaSonera 2010).
A transit arrangement is where a small ISP pays a large ISP to provide Internet
traffic exchange. The fee is generally a function of the traffic or physical connection.
Smaller ISPs generally make transit agreements with global IP carriers that can
guarantee that their Internet traffic will get routed anywhere in the world. Global
IP carriers with worldwide IP networks are often referred to as “Tier 1” carriers,
with the distinguishing characteristic that they do not generally pay any transit
fees and have the capability to reach all networks connected to the Internet (Van
der Berg 2008).
While large global carriers
from developed countries operate most of the Tier 1 networks, carriers from developing
countries are starting to emerge as significant players. India’s Tata Communications,
for example, operates a global network that makes it the world’s largest, farthest-reaching,
wholesale Internet transit provider. It provides Internet connectivity to over 150
countries across six continents, with speeds up to 10 Gbit/s.12
Tata’s reach and Internet
routing can be illustrated by running a trace route from a broadband subscriber
in Washington, DC, accessing a website in Gaborone, Botswana. Once the packet reaches
the west coast of the United States, it is turned over to Tata for delivery to Botswana
over physical connections transiting Singapore, India, and Johannesburg, South Africa
Figure 5.4 Internet Protocol Packet Route from Washington,
DC, to Gaborone, Botswana
Source: Telecommunications Management Group.
5.4.3 Implementation Issues for International Connectivity
The huge costs of deploying undersea fiber
optic and satellite networks present a challenge for many developing countries and
ISPs. Capacity on these networks tends to be owned by a few carriers, and wholesale
arrangements are not always optimum for smaller players. Likewise, a few global
IP carriers dominate wholesale access to the Internet, and smaller ISPs are forced
to pay one-way interconnection charges. Landlocked countries face special problems
since they lack coastal regions that could support a landing station for undersea
cable, while SIDSs face a connectivity challenge since they are distant from undersea
cables and lack large markets. In countries with just one physical international
link, access and pricing can become an issue, particularly if the operator of the
gateway is also a provider in other parts of the supply chain and has an incentive
to restrict competition or demand high payments.
Box 5.1 Connecting the Maldives to the International Submarine Cable Network
Source: Ibrahim and Ahmed 2008, 204.
Like most small countries, the Maldives
has been relying on satellite technology to connect to the outside world. The main
reason is the cost-effectiveness of satellite as compared to fiber cable for the
level of international traffic generated by this small country. Global submarine
optical fiber cable networks like SE-ME-WE (South East Asia–Middle East–West Europe)
have passed the Maldives, but the high cost of joining these cable consortiums prevented
the country from reaping the technical benefits of optical fiber technology.
Although satellite technology
was sufficient in the past when voice telephony was the driver of international
communications, the bandwidth consumed by data applications has surpassed the bandwidth
usage of voice applications. In 2005 the government decided that it was economically
feasible to install an optical fiber system, and a consortium was established among
three service providers: Wataniya Telecom Maldives, Focus Infocom Maldives, and
Reliance Infocom of India. The consortium, WARF Telecom International, brought the
first fiber into the country in October 2006, which connects the Maldives to the
Falcon Network at a node in Trivandrum, India. In early November Dhiraagu brought
in a cable connecting the Maldives to Colombo, Sri Lanka.
Countries are exploring
various ways to overcome the challenges of international connectivity, including
Forming public-private partnerships (PPPs) to establish direct international links.
The high cost of connecting to international networks may be insurmountable for
smaller service providers. In Kenya the government took the lead in procuring an
undersea fiber connection through an agreement to construct a cable from Kenya to
the United Arab Emirates by enlisting service providers to take a shareholding in
The East Africa Marine System (TEAMS) cable (World Bank 2011).
Establishing points of presence (POPs) in major Internet hubs. This can be cheaper
than paying transit fees. Sri Lanka Telecom established a subsidiary in Hong Kong
SAR, China, and acquired domestic and international voice and data services licenses
allowing it to offer undersea fiber optic cable capacity services from Hong Kong
SAR, China, to Asia, Europe, and North America.13
Enhancing cross-border cooperation. It is critical for landlocked countries
to coordinate and establish partnerships in order to ensure end-to-end connectivity
to undersea land stations. In Uganda the ISP Infocom leased fiber capacity from
the country’s electrical utility, allowing it to create a fiber backbone to the
Kenyan border (Kisambira 2008). From there, Infocom arranged with Kenya Data Networks
(KDN) to transport Ugandan Internet traffic to a new undersea fiber optic cable
landing in Mombasa using KDN’s national backbone (Muwanga 2009).
Improving redundancy and competition. Countries should establish varied international
connections to enhance redundancy if one link fails and to enhance competition among
international gateway operators. When different service providers offer additional
connections, wholesale international bandwidth competition also increases and prices
generally fall. Even a small country like the Maldives, where it was initially believed
that even one connection to an undersea fiber optic cable would be prohibitively
expensive, has found that an open telecommunications market with a liberal international
gateway license regime can motivate operators to invest in high-quality connectivity.
The Maldives now has two links to undersea fiber optic cable systems (Box 5.1).
Creating Internet exchange points (IXPs) close to data servers and international
By establishing its own IXP, a country can reduce expensive international traffic
by keeping local traffic local and by attracting leading global content providers.
For example, Google has a liberal peering policy and has established POPs in several
locations, including recently in South Africa at the Cape Town Internet Exchange.14