Patrice Riemens on Tue, 21 Oct 1997 19:00:56 +0200 (MET DST)


[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

<nettime> US domination of the Internet ('by default')



(This piece reaches us thanks to Michel Polman, the URL was given earlier
in nettime)

..................



                                                                    


 World Wide Weight

America's dominance of the Internet isn't just a cultural issue. It could
pose an infrastructure nightmare. 

By Andreas Evagora. 

Andreas Evagora is international editor for tele.com. He can be reached
over the Internet at aevagora@mcgraw-hill.com. 

The year is 1962. At a White House conference, aides warn President John
F. Kennedy that the newly emerging global phone network isn't really
global at all. Kennedy is told that almost every intercontinental phone
call is funneled through the United Kingdom or one of the other former
colonial powers, and that those countries control the lion's share of
international cables. Fearing that the United States will be powerless in
the new telecom era, the Kennedy administration decides to create
Intelsat, an international satellite system that would quickly end the
domination of the old powers over the global telecom network. 

Fast forward to 1997. International service providers find that the newly
emerging global Internet isn't really global at all. Reports tell them
that more than half of intra-European and intra-Asian traffic is funneled
through the United States. Once predictable traffic patterns are going
haywire, rendering established technological and economic models
irrelevant. 

Right now, there is no Intelsat on the horizon to save the day for telecom
providers outside the United States. In fact, far from globalizing, all
the signs are that Internet backbone infrastructure--and the traffic
running on it--will become even more U.S.-centric over the next few years.
This isn't just about U.S. domination: Non-U.S. service providers are
contributing greatly to the  imbalance by hatching plans to add new
capacity to the States rather than to other countries in their region.
Underpinning that trend is the concentration of content in America (more
than eight of 10 Web sites are in English) and the continued high price,
poor quality, and lack of easily available infrastructure outside the
United States. 

The increased concentration of infrastructure and traffic into the United
States not only works against the distributed philosophy of the Internet
but also threatens to endanger the 'Net's international growth and
overwork the U.S. backbone to the point of exhaustion. "The Internet today
is a hub-and-spoke system, with the U.S. at the center of the entire
world," says Neil Tagare, chairman of CTR Group Ltd. (Woodcliff Lake,
N.J.), which plans to build a new global fiber cabling network linking 175
countries. "That's not right, technologically or politically." 

Already, service providers report that Internet links to the U.S. are
straining to stand up to growing traffic volumes, while demand for other
regional and intercontinental links remains tepid. Around the globe,
U.S.-bound capacity is at a premium, with developing nations clinging onto
the 'Net by a thread. If they are lucky, Russia's 150 million people will
share 40 Mbit/s of international Internet capacity by next year. Today,
India--a nation of 900 million--has just 10 Mbit/s of Internet capacity to
the U.S.--about as much available on a LAN supporting a few dozen workers
in the United States. "The lack of non-U.S. infrastructure is holding back
real growth of the 'Net in many regions," says Petri Ojala, technical
director of the Finnish Commercial Internet Exchange (Helsinki, Finland).
"If a regional Internet community is short of capacity, it simply cannot
develop as it wants. Less capacity means less content, less innovation." 

If Internet infrastructure continues to be concentrated in the United
States, some fear that the whole non-U.S. high-tech sector will be in
grave danger. Because the 'Net is becoming a critical tool for software
development, a concentration of capacity in the States might lead to a
concentration of innovation there as well. To be sure, it's difficult to
see how India's enormous software development industry can prosper with
'Net access that is far inferior to that of competitors in the United
States. 

A few 'Net watchers warn that failure to address the Internet imbalance
could put some nations at a political disadvantage as well. "What if a new
U.S. government wanted to leverage its control of the Internet for
political ends?" asks one executive at a non-U.S. Internet service
provider. "If it wanted to embargo Cuba, it could use its influence to
control Internet communications there." 

But the outlook for regional backbones is less than bright. Such networks
are all but absent in Asia-Pacific, Latin America, and Africa, while
backbones in Europe, where the World Wide Web was invented, offer no more
than 2 Mbit/s of bandwidth (see "Un-American Activities"). Compare that
with backbone construction in the United States, where today's typical
622-Mbit/s backbones are expected to double in capacity next year. 

 Interregional Internet connections also are sorely lacking. Almost every
byte of traffic between continents passes through the United States,
hopping over at least two backbone links on the way. Concurrently,
non-U.S. network access points (NAPs), where Internet service providers
exchange traffic, are small in both number and processing capacity. "If
you were to squint at a map of the global Internet infrastructure, all
lines would roll into the U.S.," says Robert Hagens, director of Internet
engineering at MCI. "That's not a good way to build a network." 

Service providers have never quite faced anything like this before.
Exponential demand for traffic to the United States is soaring, but
investing more in these routes doesn't add significantly to the bottom
line (some phone companies that are also Internet providers offer free
local calling--and therefore free local access--to the Internet but still
need to add extra capacity), and it only increases the dependency on the
U.S. backbone. "Carriers are in a chicken-and-egg situation," says Chris
Champion, senior consultant at The Yankee Group Europe (Watford, U.K.).
"They only want to invest where there is a lot of traffic, but there won't
be enough traffic until they upgrade the backbones." 

EUNet International B.V. (Amsterdam), the European backbone operator, this
month is increasing its U.S.-bound capacity by 34 Mbit/s, to 72 Mbit/s, to
meet demand. But its intra-European backbone network, which comprises
mainly 2-Mbit/s links, faces no capacity crunch. "We don't have any
[intra-European] congestion problems whatsoever," says Wim Vink, the
company's managing director. 

The problem is, the global Internet is running headlong into an
international infrastructure regime that actually is causing congestion on
intercontinental routes. Capacity planning and demand forecasts on these
routes have been designed largely by monopolies around the predictable
needs of a staid, 5 percent a year growth in traffic. Phone companies
traditionally buy capacity on intercontintental cables 20 or 25 years
ahead of time. That's not a strategy suited to the Internet; who knows
what will happen when bandwidth-hungry voice and multimedia applications
pile onto the 'Net, as they are expected to in coming years? 

Service providers already are feeling the tremors (see "Borne in the
U.S.A."). In the summer of last year, Internet traffic from Sweden to the
United States was about half the volume of voice traffic between those two
countries. By the end of 1996, data and voice traffic volumes on the
Sweden- to-U.S. route were equal, and now data volume is double that of
voice. Meanwhile, voice still accounts for the vast majority of traffic to
neighboring Finland and Denmark. 

Traffic patterns between the world's two biggest economies also signal the
changes that lie ahead. Kokusai Denshin Denwa Co. Ltd. (KDD, Tokyo),
Japan's dominant international carrier, has 10 Mbit/s of Internet capacity
to the United States and 15 Mbit/s to the rest of Asia. Yet the ratio of
Internet traffic to the United States and to Asia is 8 to 1. "The
situation is changing very quickly," says Hiroshi Kobayashi, the carrier's
deputy director of Internet business. "Two years ago, the total ratio of
traffic flow from the U.S. to Japan was 4 to 1. Now, it is only 2 to 1."
That means Japan is sending proportionally more traffic to the United
States--a traffic shift directly attributable to the growth of the
Internet. 

 In Australia, Telstra Corp. Ltd. (Melbourne) is dealing with a U.S./Asia
Internet traffic ratio of about 6.5 to 1. "I can deal with the 1--it's the
6.5 that is the problem," jokes John Hibbard, managing director of
international carrier business. Telstra now has 130 Mbit/s of Internet
capacity to the United States, compared with 2 Mbit/s to the rest of Asia.
Still, Hibbard says of the U.S. route, "We will see a big squeeze in
1998." As a measure of just how high demand for bandwidth to the United
States has been driven, Hibbard notes that bids for the 1,000 or so
2-Mbit/s circuits on the new transpacific TPC-5 cable, due to come into
service at the end of this year, were oversubscribed by nearly four times.
"The Asia-Pacific region must reduce its dependency on the United States
to ensure that the quality of service is not dependent on the U.S. link,
which is frequently congested," Hibbard says. 

That overdependency is creating an economic as well as a technical
fallout. Non-U.S. carriers are investing in extra capacity to the United
States, without seeing returns on that investment in terms of extra
revenues. As a result, many international carriers--Telstra
included--complain that in paying the full cost of circuits to the States,
they are effectively subsidizing the U.S. Internetcommunity. They argue
that U.S. service providers should pay for at least a portion of those
circuits. 

In the traditional telephony world, international circuits are provided on
a shared-cost basis, with each carrier meeting costs to a theoretical
midpoint between two countries. Telstra and its supporters want such
principles to be looked at--although not necessarily applied fully--in
discussions about the Internet. 

This year, Telstra will lose US$10 million on providing Internet circuits
to the United States. By 2000, the total spending on U.S.-bound Internet
circuits from all non-U.S. service providers will reach US$2.5 billion,
Hibbard notes. "We are providing resources for which we are not adequately
compensated," he says. "At the same time, I am offering U.S. users access
to Australian databases without getting a brass razoo." Hibbard explains
that traffic from the United States to Australia gets a free ride, as U.S.
service providers aren't contributing to the international connection. 

IInternational service providers are lobbying U.S. regulators for
compensation. "When people first started connecting to the Internet, that
normally meant they were connecting to America, and as a result had to pay
for the connection," says Michael Behringer, senior network engineer at
Dante (Cambridge, U.K.), which operates the EuropaNet Internet backbone.
Until recently almost all Internet content has resided in the United
States, so it was fair for overseas providers to pay the bill for access
to that content, Behringer reasons. But for Internet content to become
truly global, payments should be more equitable, and non-U.S. service
providers should not have to bear 100 percent of the payment burden, he
says. 

A group of Asia-Pacific carriers, including KDD and Telstra, has already
made a proposal to the U.S. Federal Communications Commission that such
Internet circuit fees should at least be taken into consideration during
discussions on accounting rates, the fees paid by one carrier to another
for delivering an international telephone call. To give their argument on
leased lines to the United States more ammunition, several Asian carriers
recently agreed to share the cost of international leased circuits for
Internet communications between their own countries. The matter has also
been raised at meetings of the Group of Seven (G7) nations. But so far,
U.S. officials have rejected the complaint, and international officials
don't seem to hold out much hope of success. "The Americans are
sympathetic, but I doubt if they'll do anything because they don't have
to," Behringer says. 

Carriers that offer free local calls, such as Singapore Telecom Ltd. and
Telstra, are the most eager for change as they recoup little or no revenue
when users access local servers to tap the global Internet. They argue
that the Internet needs to be put on a more commercial footing. In
Singapore, 65 percent of all 'Net traffic passes through the United
States. "Because of the high cost of international bandwidth, it actually
would cost a user US$100 an hour to connect to the Internet with a
2-Mbit/s link if he was paying the real commercial rate to the U.S.," said
Tan Boon Tiong, deputy director of network technology development at
Singapore Telecom, speaking at the International Telecommunication Union's
Asia Telecom conference in June. 

Service providers point to other factors holding back the globalization of
Internet infrastructure, such as the high cost of maintenance,
installation, and operation and a lack of human resources. But many
independent Internet service providers say that old-line telecom operators
are hardly covering themselves in glory as the 'Net spreads its tentacles.
Independent service providers and builders of backbones say that high
leased-line costs remain the single biggest factor holding back the
expansion of regional backbones that could help keep more Internet
communications away from the United States. Lack of competition and a
scarcity of infrastructure, particularly cross-border links, continue to
keep leased-line prices in Europe three to 10 times more expensive than
equivalent lines in the United States. Service providers say that they
must wait up to five months for a cross-border E3 (34-Mbit/s) line, where
such service exists. 

Ironically, it's the traditional telcos, which set those high prices, that
are gradually taking over the operation of pan-European Internet
backbones. "European carriers are schizophrenic," concedes one official at
an international carrier alliance. "They will cry about lack of
liberalization in another country but do all they can to delay
liberalization in their own market as long as possible. That's human
nature." 

That human nature is only serving to encourage the flow of 'Net traffic
and facilities to the United States. "Dante's backbone took 18 months to
organize and two years to turn into a reality," Behringer says. "That
delay was far too long and was mostly caused by arguing with telcos." 

It's not just high leased-line prices that anger independent ISPs.
Deutsche Telekom, now Europe's largest ISP, still refuses to peer its
network at Frankfurt's commercial NAP, where more than 25 independent
German ISPs exchange traffic. Instead, Deutsche Telekom maintains a single
peering deal with Deutsches Forschungsnetz (DFN), a scientific and
academic network. That lack of domestic peering means that all
communications between Deutsche Telekom and other German ISPs must be
routed via the United States. 

"Peering via the U.S. affects Deutsche Telekom customers and our customers
because of the delays on U.S. lines, which are normally saturated," says
Thomas Bastian, senior technical director at CERFnet Germany Inc.
(Frankfurt), the German unit of U.S. ISP TCG CERFnet (San Diego). In
response, he says, all independent ISPs have announced that they will
cancel any private peerings with DFN as of this coming January. 

"We are confident that things will change, but everything takes so long in
Europe," Bastian says. "Our U.S. operation connects into NAPs at 155
Mbit/s, but here in Germany it's at 2 Mbit/s. It's very expensive just to
set up a system, and those costs are inevitably passed on to users, which
slows market development." 

Until more capacity is made available, there's not much chance that the
global Internet's dependency on the United States will fade away. CTR
Group's Tagare, a former executive at Nynex Network Systems (Bermuda)
Ltd.--one of the founders of the FiberOptic Link Around the Globe (FLAG)
project building a new global undersea cable--says that a lot of capacity
remains warehoused to block competition. "Active capacity is no more than
50 to 60 percent of the bandwidth out there--the rest is warehoused," he
says. "The people who need capacity desperately can't buy it because none
is available, while those that have it can't use it because they don't
have sufficient demand for it. That's an extremely inefficient model." 

WorldCom International Inc. (New York), another company planning to lay
new international cabling, also places much of the blame for the current
malaise at the door of incumbent telcos, in particular for failing to spot
the tide of demand for data communications. WorldCom International says
that voice today accounts for 80 percent of international traffic, but
will make up only 20 percent by the early part of next decade. 

"Terabyte requirements will soon be upon us, but carriers got caught
asleep making cozy, 4 percent growth demand forecasts," says Colin
Williams, the company's international executive vice president. "Today, 90
percent of Internet traffic goes to the U.S., but we haven't yet scratched
the surface of bandwidth requirements across the Atlantic." 

Translation: Don't expect a truly global Internet anytime soon.


---
#  distributed via nettime-l : no commercial use without permission
#  <nettime> is a closed moderated mailinglist for net criticism,
#  collaborative text filtering and cultural politics of the nets
#  more info: majordomo@icf.de and "info nettime" in the msg body
#  URL: http://www.desk.nl/~nettime/  contact: nettime-owner@icf.de