Are poor countries
subsidizing the rich?
Framing the issues
|
Photos.com |
When an international telephone call
is routed
from one country to an
other, the operator in the country
that originates the call has traditionally made
a compensatory payment to the operator in the
country that terminates the call. These payments usually cancel
each other out, if traffic is balanced,
but they become more significant when traffic in one direction is higher than that
in the return direction. This
system — also called the “half-circuit model” — was traditionally
based on “accounting rates” that
were negotiated bilaterally. ITU estimates that, between 1993 and 1998
when the uneven pace of telecommunications liberalization generated
large traffic imbalances, net flows of
settlement payments from developed countries to developing ones
amounted to some USD 40 billion. But since the late 1990s, these payments have declined and, as a greater
share of traffic shifts to the Internet,
may even have reversed.
The high cost of Internet
connectivity and bandwidth
in developing countries
For international Internet charging, the system is quite different, based on a so-called
“full-circuit” model. According to a discussion
paper published in January 2005 by ITU and
the International Development Research Centre
(IDRC): “Developing countries wishing to connect to the global Internet backbone must pay
for the full costs of the international leased line
to the country providing the hub. More than 90
per cent of international IP connectivity passes
through North America. Once a leased line
is established, traffic
passes in both directions, benefiting the
customers in the hub
country as well as the
developing country,
though the costs are
primarily borne by the
latter. These higher
costs are passed on to
customers in developing countries. On the
Internet, the net cash
flow is from the developing South to the developed North.”1
The authors of the
paper argue that the
high cost of Internet connectivity and bandwidth inhibits the growth
of Internet usage in much of the developing
world, especially the least developed countries
(LDC). One reason for this high cost is that most
developing countries use international bandwidth to exchange data at a local level. “When
an African Internet user sends a message to a
friend in the same city or a nearby country, that
data travels all the way to London or New York
before going back to that city or the nearby country.” It has been estimated that this use of international bandwidth for national or regional data
costs Africa, for example, some USD 400 million
a year. When the North-South imbalance in telephone traffic was no longer acceptable by the
North, a reform was launched to redress it. Now
there is a similar imbalance for Internet traffic.
However, Internet backbone providers in the
developed world respond that they do not
charge developing-country Internet service
providers (ISP) any more than their other customers.
They believe that the majority of international costs are incurred for a number of reasons, such as poor telecommunication infrastructure at national and regional levels, fewer
peering and exchange points than elsewhere,
and a genuine lack of competition in
many developing countries.
Still much work needed to improve
international Internet bandwidth
Inter-regional Internet bandwidth flows, 2003
|
|
Source: PriMetrica. |
These problems afflicting small developing economies have been the
subject of research in a number of ITU
Internet Case Studies (see http://www.itu.int/ITU-D/ict/cs/). For most African countries, for example, the international gateway that would be used to
carry data to other African countries remains in the hands of monopolies with
no competition on rates. Consequently,
prices remain artificially high. The good
news is that this is beginning to change
as exclusivities granted to incumbents
in many countries come to an end —
and as these countries revise their competition frameworks.
Uganda, as one of the early adopters
of innovative communication policies in
Africa, provides a valuable example of
what a competitive market can help
achieve. Since the liberalization of the
Ugandan telecommunication market in
1997, there has been a dramatic growth in
Internet users with estimates increasing from
less than 5000 in 1996 to more than 125 000 at
year-end 2003, according to ITU indicators. In
Nepal, prices dropped to the lowest level in the
South Asia region, when the country liberalized its market for very small aperture terminals (VSAT) in 2000.
If the rapid uptake of mobile phones is anything to go by, developing-country users have
demonstrated a willingness to pay for information and communication technology (ICT) services when provided in a way they can afford.
However, some of the problems are structural
and largely unavoidable, such as the low level
of demand in LDCs and Small Island Developing States, which tends to push up unit costs.
Some countries and regions are not served by
undersea cables, and have to rely on highpriced satellite access. There is also a nagging
suspicion that the international bandwidth
market is not as competitive as might be expected, especially since the consolidation of recent years. Thus, there is evidence of market
failure that liberal telecommunication policies
cannot fully address.
|
ITU 050016/Photos.com |
The peering/transit dichotomy
|
ITU 050015/Photos.com |
If developing countries had a greater ability
to exchange traffic locally at a national level
and regionally, they would not be paying for expensive international bandwidth for their
connections. Similarly, if these countries had
more outgoing traffic and more regional carriers, these carriers would be able to peer with
their international counterparts and lower the
costs of international bandwidth (see box on
“What is peering”?).
An Internet exchange point (IXP) interconnects ISPs in a country or region, allowing them
to exchange domestic Internet traffic locally
without having to send that traffic (for example,
e-mail messages or Web traffic) across several
international hops to reach their destination.
But, the reality today is that most developing
countries lack local IXPs.
For now, the most promising option for most
developing-country ISPs to connect to the global Internet is via a transit
agreement signed with
their upstream providers. However, because developing-country ISPs have a
small customer base, the
international Tier-1 and
Tier-2 providers have no
business incentive to enter
shared-cost peering agreements with them. As a result, these ISPs have to
bear the full costs of both
outbound and inbound
traffic exchange under the terms of the transit
agreement, in addition to the leased line costs.
The ISP on the other end of the international link
does not share the cost of exchanged traffic. This
means that the developing-country ISP has to
pay 100 per cent of the international transit costs
for all packet traffic (e-mail, Web pages, file transfers and so on) that originates and terminates
with its customers.
Take the example of a customer of a
Mozambican ISP. When that customer sends an
e-mail message to a friend in the United States,
it is the Mozambican ISP which bears the full
cost of the packets’ outbound transmission over
its international link. Neither the recipient’s ISP nor intermediate upstream carriers
bear any of the transit cost. Conversely, when that friend in the
United States replies back by e-mail
to Mozambique, or makes a voice
call routed over the Internet, it is
still the Mozambican ISP who again
bears the full cost of inbound transmission over its international link2. And so, in the process, the Mozambican ISP’s
customer bears the brunt by paying higher subscriptions.
What is peering?
Peering refers to a relationship between two or more ISPs of similar
size, in which the ISPs create direct links between each other and agree
to forward each other’s packets directly across this link. Imagine, for
example, that a client of ISP “X” wants to access a website hosted by
ISP “Y”. If X and Y have a peering relationship, the HTTP packets will
travel directly between the two ISPs. In general, this results in faster
access since there are fewer hops. And for the ISPs, it is more economical because they do not need to pay fees to a third-party network
service provider (NSP). Peering between equals is equivalent to
“sender-keeps-all” in the half-circuit model.
Peering can also involve more than two ISPs, in which case all traffic
destined for any of the ISPs is first routed to a central exchange, called a
peering point, and then forwarded to the final destination. In a regional
area, some ISPs exchange local peering arrangements instead of, or in
addition to, peering with a backbone ISP. In some cases, peering charges
include transit charges, or the actual line access charge to the larger
network.
However, when ISPs are of unequal size or negotiating ability, or
where types of traffic vary, peering is generally replaced by commercial agreements in which smaller ISPs must pay for traffic exchange.
Sources: Webopedia, Whatis.com (Extract adapted from Via Africa: Creating
local and regional IXPs to save money and bandwidth.) |
Internet exchange points:
A possible way forward
In Asia, IXPs have been established since 1996
in a number of the more developed countries
across the region. The largest IXPs are in Seoul
(Republic of Korea); Tokyo (Japan); Perth (Australia); Singapore; Wellington (New Zealand);
and Hong Kong, China. But developing countries,
too, in Asia-Pacific are now playing catch-up, and
the number of IXPs is growing in Cambodia,
Mongolia (see box) and Nepal.
For Africa, more continental interconnection
would enable African ISPs to aggregate intra-African traffic and negotiate
better transit prices from the
global backbone providers. With
the exception of countries of
the Southern African Development Community (SADC),
there are very few intercountry links and only a minority of African countries
are linked by fibre. Even
where fibre exists, as in the
case of the SAT-3 cable, the international gateway of a particular country may be in the
hands of a monopoly incumbent
telecommunication company,
with no competitive pressure on
prices.
The African Internet Service
Provider Association (AfrISPA)
has played a key role in setting
up IXPs with support from a variety of public and private partners. In October 2002, AfrISPA
published a policy paper entitled
“Halfway Proposition”. The paper pointed to the high cost of
international bandwidth as one
of the causes of high prices for
African Internet users. AfrISPA
proposed the creation of traffic
aggregation within Africa as an
approach that would help avoid
the need to invest in expensive additional capacity between Europe or North America and
African countries. By September 2004, there
were ten national IXPs in Africa: Democratic
Republic of Congo, Egypt, Kenya, Mozambique,
Nigeria, Rwanda, South Africa, Tanzania,
Uganda and Zimbabwe. There are no IXPs in
French-speaking West Africa today, but talks
are under way.
While examples have focused on the challenges and opportunities for IXP deployment
in Africa and Asia, the lessons and strategies
are relevant to the rest of the developing world.
The ITU/IDRC paper concludes that if the level
of traffic is high enough to be exchanged locally, then the IXP concept represents a rational
solution for the developing countries.
Mongolia’s successful IXP
|
Photos.com |
The case of Mongolia demonstrates that a
combination of ISP cooperation and at least
tacit support from governmental authorities
can lead to the rapid and successful establishment of an IXP in a developing country.
In January 2001, a group of leading Mongolian ISPs met in Ulaanbaatar to explore the
creation of a national IXP. At the time, all
Mongolian ISPs were interconnected via Tier-1 and Tier-2 providers in the United States or Hong Kong, China. As a result, satellite
latencies amounted to a minimum of 650 milliseconds (or over half a
second) for each packet of data in each direction. And so, costs were
unnecessarily high. Not surprisingly, few Mongolian Internet business services were hosted within the country.
Within three months, these leading ISPs were able to complete planning for an independent exchange. In April 2001, the Mongolian
Internet Exchange (MIX) was launched with three ISP members. By
March 2002, MIX had six ISP members, with traffic increasing steadily
between them. Today, local latency is less than 10 milliseconds per
transaction (compared with a minimum of 1300 milliseconds in the
pre-MIX days), and an average of 377 gigabytes of data are transferred domestically each day among MIX’s members. Moreover,
each transaction that is exchanged domestically frees up an equal
amount of international bandwidth, improving connection speeds
and reducing latency over Mongolia’s international links.
[Source: MIX.] Extract adapted from “Internet Exchange Points: Their
Importance to Development of the Internet and Strategies for their
Deployment — The African Experience”, a paper of the Global Internet
Policy Initiative.
|
1 Via Africa: Creating local and regional IXPs to save
money and bandwidth — Discussion paper prepared
for ITU and the International Development Research
Centre (IDRC) for the 2004 Global Symposium for
Regulators. The paper was written by Rusell
Southwood, CEO, Balancing Act, and was released
in January 2005 as a joint publication of ITU and
IDRC, following a comment period that closed on 30
December 2004. 2 Internet Exchange Points: Their
Importance to Development of the Internet and Strategies for their
Deployment — The African Experience, a paper of the Global Internet Policy
Initiative. |
|