[Forward from RRE ... John]

============================
From: Phil Agre [mailto:[log in to unmask]]
Sent: Friday, January 19, 2001 11:26 PM
To: Red Rock Eater News Service
Subject: [RRE]Telegeography 2001



=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
This message was forwarded through the Red Rock Eater News Service (RRE).
You are welcome to send the message along to others but please do not use
the "redirect" option.  For information about RRE, including instructions
for (un)subscribing, see http://dlis.gseis.ucla.edu/people/pagre/rre.html
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

Date: Fri, 19 Jan 2001 18:20:13 -0800
From: Jessica Marantz <[log in to unmask]>

[...]

TELEGEOGRAPHY 2001:
GLOBAL TELECOMMUNICATIONS TRAFFIC STATISTICS AND COMMENTARY

by TeleGeography, Inc.
Washington, DC: TeleGeography, 2000
ISBN 1-886142-26-2

http://www.telegeography.com/Publications/tg01.html

TABLE OF CONTENTS

List of Figures
List of Country Traffic Tables
Preface
Executive Summary

Carriers
  The Growth of International Services Competition
  Market Shares of International Carriers
  The Top 40 International Carriers
  The Top U.S. International Carriers
  A Primer on Bandwidth Exchanges

Pricing
  Overview of International Pricing Trends
  International Carrier Call Costs from/to the U.S.
  International Private Line Prices
  Settlement Rates and Benchmarks
  National Interconnection Rates and a Primer on Interconnection Accounting
  Retail Prices and Trends
  Retail and Wholesale Rates:  PSTN versus VoIP
  Follow the Money:  Network-to-Network Payments for Internet
Telephony and Other IP Traffic Streams

Facilities
  MANs:  The Golden Mile
  Submarine Cable Systems
  International Circuit Usage by U.S. Carriers
  International Communications Satellites

Internet Backbones
  International Internet Backbones

Traffic Analysis
  Overview of International Traffic Trends
  Bypass and Refile Traffic
  VoIP Routes & Traffic
  International Traffic from Mobile Phones
  International Call Quality Metrics

Traffic Summary
  Global Traffic Review
  International Traffic by Region
  International Traffic by Country

Global Reference:  Blue Pages


INTERNATIONAL INTERNET BACKBONES

What is an Internet backbone?  And when is it international?  The
questions are not as straightforward as they might seem.  International
Internet backbones are private data links which cross international
political borders, run the Internet Protocol (IP), are reachable
from other parts of the Internet, and carry general Internet traffic:
e-mail, Web pages, and most of the other popular services which have
come to define today's Internet.

That means that international IP links devoted to just one type
of traffic -- notably, Voice-over-IP (VoIP) -- are excluded from our
definition of backbones on the public Internet.  If VoIP is excluded,
though, then why publish international Internet backbone data in a
book on international telephony?  The answer: because it just might be
important.

Despite a history stretching back more than 30 years, today's Internet
really began its push toward ubiquity during the 1990s in a rapid
transition from academic network to commercial networks.  What evolved
was a decentralized infrastructure whose end-to-end design made it
possible for users to create new network applications without asking
too many people's permission.

The resulting infrastructure took media services based on text
and simple graphics and turned them into the most widespread media
platform since television.  That ubiquity only fueled its popularity,
however, and soon people were stuffing two-way voice telephony,
streaming video, and other bandwidth-intensive applications into
the public Internet.  They did this not because the Internet's
then-infrastructure was particularly well-suited to such services,
but because running them over the Internet meant bringing together
multiple services on a single platform.  On networks, the whole is
always more than the sum of its parts.

The so-called "public Internet" is at a crossroads.  How will it
accommodate very different types of traffic inside the same networks?
Some want to solve the problem by bestowing Quality of Service (QoS)
provisions upon IP so that networks can distinguish between what needs
to be delivered immediately and what needs to be delivered with care.
Some, pressed for time, prefer to forego fancy traffic engineering
by throwing more bandwidth at the problem, hoping to give every
packet enough room to get to its destination in style.  And some
are abandoning the public Internet altogether: distinct backbones
are emerging for self-similar traffic generators, like VoIP or the
Usenet's text- and photo-oriented discussion groups.  Those highly
specialized, single-service backbones are not included here.

That sharpens the scope of what we mean by "Internet" backbones.
But it doesn't close off their possibilities.  There is increasing
excitement over a "new public network" infrastructure which meshes
PSTN (public switched telephone network) and IP infrastructure into
the backbone of tomorrow's public communications facilities.  If
the feverish activity taking place around the world can successfully
achieve the economies of scale and creative possibilities that
interoperable communications services represent, these backbones will
have to come together to look like the Internet as many engineers have
always drawn it -- a cloud.

Bandwidth, Not Traffic

The maps and statistics on the following pages show international
Internet backbone capacity, or bandwidth -- not traffic.  There are
several reasons to keep track of international Internet bandwidth.
One is to provide a rough metric for matching supply and demand.
But there is another reason: bandwidth take-up may provide a clue
to Internet traffic statistics, which are still in very short supply.

International Internet bandwidth is growing faster than international
Internet traffic, however.  In the past few years, tremendous physical
infrastructure builds began to come on-line.  Because raw bandwidth
does not translate immediately into Internet capacity, however -- it
must first be lit, sold, deployed, and integrated into data network
operations -- the numbers showed what, to casual observers, appeared
to be a mismatch between physical capacity and Internet capacity.
In 2000, however, bandwidth started diffusing up the network layers,
moving from physical installation to actually existing network
services.  As Internet capacity began to take advantage of the fiber
explosion, two-and-a-half gigabit per second OC-48 (STM-16) links
running Internet Protocol became, if not common, at least widespread.

All this new Internet capacity makes network bandwidth less useful as
a proxy for traffic.  But it does provide important insights as to how
traffic is routed.  Historically, steep intraregional bandwidth costs,
a comparative lack of local content, and limited regional coordination
had caused the U.S. to become the Internet's central switching office,
even for data flows within a region.  Last year, we found that the
Internet was still U.S.-centric but that places like Western Europe
and, to a lesser extent, eastern Asia were beginning to develop as
secondary hubs (see Figure 1.  Interregional Internet Bandwidth,
2000).

In 2000, this trend continued.  In Europe, international capacity
between European countries again grew at a noticeably faster clip
than did outgoing international bandwidth, nudging the continent's
in-region connectivity to beyond 75 percent of its total international
Internet bandwidth.  Asia's intraregional connectivity, too, grew
more than twice as fast as to anywhere else, doubling the proportion
remaining within the region to one-sixth of total international
Internet bandwidth.

Enormous differences still remain from country to country. The
U.S.-centric pattern wanes only with substantial and sustained
infrastructure builds of the sort that has swept Europe, rolled
into Asia, announced itself in Latin America, and stalled in most
of Africa.  The connectivity divide is reflected in the larger
pattern of global net bandwidth (see Figure 2.  A Question of Scale).
Although fiber builds in Latin America and the Caribbean increased
substantially in the last year, the Internet had not yet caught up
as of mid-2000, and the continued reliance on slower, more expensive
satellite links for international connectivity translated into another
year of the World Wide Wait.  In continental Africa, similarly,
the number of countries connected at above 10 Mbps went from three
to six -- but Africa-U.S. Internet bandwidth remained the continent's
fastest-growing route.

The Players

Approximately 300 International Internet Service Providers (IISPs)
own, lease, or otherwise get hold of transborder network capacity;
place routing computers at either end; and use these segments to
cobble together logical networks that, together, form the Internet's
international backbone.  Three hundred may seem like a lot.  Not
all backbones are equal, however: in mid-2000 the ten largest IISPs
controlled three-quarters of international Internet bandwidth.

Some observers try to make sense of the Internet's snarl of networks
by dividing them into three or four tiers.  Under that framework,
"Tier Ones" are the handful of global backbone operators who have rich
interconnection relationships with all other significant providers;
"Tier Twos" are the not-quite-Tier-One backbones who end up having to
pay for some of their direct backbone connectivity; and "Tier Threes"
and "Fours" are the national/regional and local ISPs, depending on the
context and topology, in question.

Those definitions are somewhat fuzzy -- and for good reason.  As
a whole, the Internet service provider world is not segmented into
hierarchical divisions, so hard-and-fast typologies just aren't
possible on a global scale.  The same is true, perhaps more so, of the
IISP segment.  Instead, we have identified four groupings around which
IISPs cluster.  However imperfect, these markings on the IISP spectrum
help understand which way they lean:

Global IISP.  Two kinds of IISPs engage in activities which place
them in the "global" range.  One is a set of very large players who
have strong historical roots in the U.S. Internet, either in origin
(AT&T/Concert, Genuity, WorldCom, PSINet, Sprint) or by osmosis; Cable
& Wireless acquired much of MCI's Internet backbone as a result of
the MCI-WorldCom merger.  The other group of IISPs are providers who
feature managed IP bandwidth over bent-pipe (point-to-point) satellite
as important parts of their services portfolios, typically Intelsat
signatories like Telecom Italia, whose Seabone offering connects many
countries around the world, or service providers like Interpacket.
The effort to move from the second set of global IISPs into the first
-- Teleglobe attempted this during the late 1990s -- is a key dynamic
within this grouping.

Regional IISP.  A regional IISP specializes in operating backbone
connectivity between different countries in a single region, like GTS
E-Bone in Europe, Pacific Century CyberWorks or Telstra in Asia, and
Africa Online.  Because of the impressive build-out in Europe during
the past two years, Western Europe is probably the best example of
the impact that regional IISPs can have on reconfiguring a single
region's topology map.  At the other end of the spectrum, a number of
IISPs continue to vault into the regional area by purchasing existing
smaller, nationally-based IISP networks; once, these were dominated
by former incumbents, but an increasing number of new entrants have
borrowed this strategy as well, fuelled by the international spread of
venture capital and Initial Public Offerings.

National IISP.  Typically, this is an Internet provider which has
acquired international connectivity as part of a national or local
service; which acts increasingly as an upstream provider for other
providers who have little or no international connectivity; and which
moves to expand into neighboring countries.  In 2000, this sector
actually shrunk, as existing players federated or were bought up to
form regional IISPs.

Academic IISP.  Research networks, including those operated by
academic institutions, often act as international connectivity
providers alongside commercial IISPs.  In many environments, they
operate high-capacity, leading-edge systems, catalyzing Internet
development -- examples are Europe's DANTE TEN-155 and GIANT projects
-- but they are increasingly specializing in exclusively academic
and research traffic as part of the international coordination of
Internet2 and advanced research applications.

Methodology

The data depicted on the following pages -- using different scales for
different regions -- result from a TeleGeography, Inc. study completed
in October 2000.  The research focused on the network topologies
of over 300 IISPs operating international Internet links -- routers
or switches directly connected across an international border over an
internal network.  These links and their capacities were then tracked
through over 300 cities in more than 180 countries.  Each IISP's
network routes and capacities were derived from a combination of
public documents, confidential interviews, and computer-based network
analysis tools.

The study grouped specific switch and router locations according to
Consolidated Metropolitan Statistical Area, Census Metropolitan Area,
or the equivalent.  Only the IP network was mapped, instead of the
physical network infrastructure which runs beneath it.  In cases where
IISPs had provisioned relatively new dedicated IP capacity, the study
did not include the capacity unless it was believed to be operational
and available for public Internet traffic as of mid-2000 (i.e.,
bandwidth kept in reserve was excluded).  A final note: due to the
complex and ever-changing nature of network architectures, omissions
may have occurred.


A PRIMER ON BANDWIDTH EXCHANGES

The term "bandwidth exchange" has been pulled and stretched to cover
the canvas of diverse business models and operational approaches.
But what exactly is a bandwidth exchange?  Part of the definitional
problem originates from the complexity inherent in the telecom
industry's web of supply and demand -- whether it be for cross-border
telephone calls, intracity fiber optic connections, or access to
the Internet's cloud.  A second, more obvious, obstacle to a clear
definition is the newness of the bandwidth exchange business --
many companies have redirected and redefined their strategic focus
(and marketing programs) more than once in just a few short years of
existence.

Nonetheless, these bandwidth bazaars all share a common purpose: to
facilitate transactions between buyers and sellers.  Their challenge
-- and the source of their diversity -- is to develop the best
methodology for facilitating each type of transaction.  This primer
differentiates the services exchanged, as well as the role played by
the matchmaker in each deal.  We will begin with a brief description
of the companies involved -- including Band-X, now TeleGeography's
corporate parent.

Background

The traditional process of buying and selling communications
bandwidth -- for carriers, ISPs, and multinational corporations alike
-- can be time -- consuming and labor-intensive.  The process typically
requires direct negotiation over price, quality, and delivery.  An
exchange, however, can extend, complement, or replace all or part of
a buyer's or seller's sales force at various stages of the process.
Furthermore, bandwidth buyers can use exchanges to find quickly the
best price/quality ratio on offer, and bandwidth sellers with excess
capacity can earn incremental revenues with minimal effort.

A bandwidth exchange may consist of a bullpen of brokers, perhaps
part of a larger team of traders, who spend their days scanning price
listings and phoning potentially interested parties.  We classify
these brokerages, along with bulletin-board operators, as "virtual
matchmakers".  Alternatively, an exchange may be based upon a switch
connected to a computerized system where anonymous buyers and sellers
swap traffic.  Such exchanges, which have facilities where members
interconnect their networks for physical delivery, fall under the
category of "physical matchmakers".  Both operational models, virtual
and physical, assist buyers and sellers of bandwidth in finding
counterparties and completing transactions.  Of the more than 35
companies with actual or stated plans to trade bandwidth, at least
twelve have facilities which route capacity between buyers and
sellers.  The remaining two-thirds of existing exchanges, accordingly,
are virtual matchmakers.

Bandwidth exchanges also differ by their own degree of involvement
in the bandwidth transaction.  Some exchanges are neutral, favoring
neither buyer nor seller; others are created by a party to the trades.
The latter category includes "market maker" exchanges -- exchanges
whose founders are in the carrier or capacity building business.
The oft-cited market maker example is Enron, the energy company which
is investing heavily in both a nationwide network build-out and the
development of a bandwidth trading exchange (see Figure 1. Selected
Bandwidth Exchanges).

But why would an energy company enter the world of telecommunications
bandwidth?  The power industry's experience with energy trading may
prove applicable to the emerging bandwidth marketplace.  Enron's
vision of the communications market relies on the commoditization
of bandwidth, where a liquid market allows the trading of forward
contracts and financial derivatives by bandwidth users, as well
as speculators and arbitrageurs.  This vision may bear fruit.
Our discussion here, however, focuses on the current state of the
bandwidth exchange industry and emphasizes the delivery market for
bandwidth -- the buying and selling of bandwidth for actual use --
rather than the still mostly theoretical trading world of bandwidth
financial instruments.

Services

To date, most bandwidth trading has focused on international telephone
calls, or "minutes," and raw network connectivity, or "bandwidth
circuits."  More recently, bandwidth exchange Band-X has also begun
trading Internet transit, or "routed IP" services.  Some exchanges may
also broker related services to their customers, such as colocation
space in carrier hotels, or empty ducts between points in a city.

Minutes.  Telephone carriers typically meter service to their
customers in minutes or a portion thereof.  Thus, because
carriers around the world have agreed on standard definitions of a
"conversation minute," they are able to route minutes easily from
one network to another through their switches.

Many exchanges, including Band-X, Arbinet, and TheGTX, currently trade
minutes.  The total volume of minutes traded through exchanges is
difficult to ascertain.  But our research indicates that international
traffic flows through exchanges could reach 300 to 500 million minutes
in 2000 -- about 0.5 percent of the world's traffic.

The minutes bought and sold through an exchange may include either
traditional, circuit-switched minutes, or Voice-over-IP (VoIP)
minutes.  Some exchanges specialize in IP minutes (e.g., Pulver.com's
Min-X); others allow members to specify if VoIP minutes are desired
(e.g., Arbinet).  Those exchanges with their own switching facilities
may trade circuit-switched and IP minutes without differentiation if
their switches route both SS7 (circuit-switched) signaling and IP
protocols.  Although hard numbers are difficult to come by, it is
generally agreed that VoIP minutes make up a small, though growing,
share of the total minutes traded.  Arbinet, for example, reports that
nearly ten percent of its traded traffic is Voice-over-IP.

Minutes offers are typically listed on exchanges on a country-to-
country or city-to-city basis: for example, "U.S. to India at $0.13
per minute."  And although each minute is, by definition, the same 60
seconds of connect time, a minute's quality is not standard and can
vary considerably.  For this reason, minutes offers frequently contain
a quality metric as well (see Figure 2.  Common Bandwidth Trading
Parameters).

Bandwidth Circuits.  International carriers of voice and fax services
may purchase wholesale minutes, as discussed above, to route calls
to their desired destinations.  Alternatively, buyers may lease or
purchase bandwidth circuits between points and provision switches
at the ends.  Thus, minutes and bandwidth circuits can be, in some
instances, substitutes for each other.  But there the comparison ends.
Minutes are a service which ride on a physical circuit; bandwidth
circuits are the capacity which may be provisioned to carry any
application.  A buyer purchases minutes by quantity, and bandwidth
circuits by capacity, for a given time period.

Bandwidth circuits are typically listed in city pairs and by potential
carrying capacity: for example, New York to London at 2 Mbps.  In
addition to the geographic and capacity parameters, bandwidth circuit
exchangers must specify their commitment period.  Commitments can
range from one year leases to Indefeasible Rights of Use (IRUs) for
the lifetime of the facility.

Bandwidth circuits take various forms.  The circuit may be a satellite
link, a segment of a terrestrial network, or an undersea submarine
cable connection between world regions.  As the medium used to carry
the traffic differs, so too does the circuit's level of provisioning.
A company may purchase dark fiber, which is optical fiber not
connected to transmission equipment.  Alternatively, a company may
prefer a circuit which is already "lit" to handle its application
of choice.  A new class of bandwidth products -- wavelengths --
has recently emerged for buyers and sellers of optical fiber.
A wavelength, or a single channel on an optical fiber system, is
typically sold at 2.5 Gbps or 10 Gbps increments (see TeleGeography's
International Bandwidth 2000 report for a detailed explanation of the
various options available).

Given the wide range of bandwidth circuit increments and technologies,
determining universal standards and contracts is far from a simple
task.  Band-X has developed a standardized Service Level Agreement
(SLA), which includes both provisioning time and SLA post-
provisioning, for bandwidth circuit trades on its exchange.  A few
organizations, including one organized by carrier association CompTel,
have convened to resolve the issue; not surprisingly, one of the
thorniest issues has been the liquidation of damages if one party
does not abide by the terms of a bandwidth deal.  As with creating
quality of service measurements for bandwidth circuits, standardizing
bandwidth contracts across multiple exchanges and carriers will take
time.

How active is bandwidth circuit trading?  As the pooling points
necessary for facilities-based bandwidth circuit trading are still
in their early stages of development, the number of deals facilitated
by virtual matchmakers far outnumber those by physical matchmakers.
From July to September 2000, brokerage Chapel Hill Broadband reported
two to three hundred requests for wholesale capacity through the
Bandwidth.com lead-generating website.  Physical matchmaker Enron, in
contrast, says it will facilitate about 100 bandwidth deals this year.

Internet transit (IP routed).  Exchanges may also help match buyers
and sellers of upstream Internet access, or transit.  To date, Band-X
remains the only bandwidth exchange to provide physical matchmaker
services for Internet transit.  Like bandwidth circuit buyers,
Internet transit buyers purchase dedicated capacity for a specific
time period.  However, unlike bandwidth circuits, Internet transit
capacity is not restricted to a point-to-point route.  Instead, a
buyer of Internet transit receives a guaranteed connection to an
Internet backbone, which will carry the buyer's traffic over various,
unspecified paths to reach its intended recipient.  Recipient
locations are IP addresses, not physical places.

The exchange itself does not provide the IP transit; rather, the
exchange helps the buyer to find and connect to an "upstream" ISP.
For example, a large company may require high-speed Internet access
for the next few months.  Instead of negotiating with a number of
ISPs, determining the best price and quality match, and signing a
contract with the chosen provider for a fixed length of time, the
company may enlist the services of a bandwidth exchange.  Buyers can
choose from various, anonymous ISPs based on price and performance,
and, in some cases, can switch providers as often as once a month
(for a detailed description of Band-X's routed IP service, see Figure
4. Case Study: Band-X).

Operational models

The business practices of bandwidth exchanges are in constant flux,
and many include elements of multiple business models.  Keeping this
in mind, we consider two operational categories: virtual and physical
matchmakers.  The examples highlighted in the accompanying figures
illustrate the individual approaches of certain exchanges.

Virtual matchmaker. Virtual matchmakers help to put bandwidth
buyers and sellers together without physically interconnecting the
counterparties.  There are two non-exclusive categories of virtual
matchmakers: electronic bulletin-board operators and over-the-counter
(OTC) brokers.

Electronic bulletin-board operators post sellers' offers and buyers'
bids on a website, speeding the information-gathering process for
the parties involved.  The posting process may be automated, with
website members directly listing their prices; or, employees of
the bulletin-board operator may enter the information into the
site after communicating with the buyers and sellers.  Generally, a
bulletin-board offers a passive means of generating leads, as it is
left to the interested parties to act on the bids and offers posted
(see Figure 3. Case study: Bandwidth.com).  Band-X, a virtual
matchmaker for bandwidth circuits, has developed a more interactive
approach by holding reverse auctions on its website, where sellers
place competitive offers on specified routes in real time.

Frequently referred to as "dating services," OTC brokers search
for bandwidth or minutes terms which match buyers' or sellers'
requirements; the broker often works out the details of the deal
over the phone.  An OTC broker may rely on electronic bulletin-board
services to find prices, and then call clients to alert them to
attractive deals.  Or a broker may have his own "inventory" of
available bandwidth that he has collected, as an independent operator,
from contacting sellers of capacity.  In contrast to bulletin-board
operators, OTC brokers provide active matching services to buyers and
sellers, often adding value to their clients through their personal
connections and effort.

OTC brokers are generally carrier-neutral-their concern is to complete
bandwidth deals, not to favor one buyer or seller over another.
At times, however, a broker may represent one party from the bid
or offer stage through to contract negotiation.  For example, Chapel
Hill Broadband, which most often acts as a neutral broker, may also
be hired to negotiate a specific transaction on behalf of the buyer
or seller.

Virtual matchmakers, in addition, may facilitate the delivery of
the minutes or bandwidth trade.  For example, although by definition
a virtual matchmaker owns no interconnect facilities, it may hold
information on where its clients are colocated, and suggest locations
for interconnection.  As with other deals made through virtual
matchmakers, these transactions are not anonymous, as both parties
must work together to provision their networks for delivery.

Physical matchmaker. Physical matchmakers do more than match buyers to
sellers; they actively facilitate the delivery of the bandwidth deal
through their own facilities.  Such exchanges deploy switches or hubs
where their members may interconnect.  The decision of which switch
or hub equipment to use depends upon the kind of services the exchange
provides; facilities that carry minutes differ from facilities that
allow bandwidth circuit trades.  For minutes, switches of various
shapes and sizes route calls from seller to buyer and monitor their
flow.  Bandwidth circuits, in contrast, are not so easily switched
between carriers.  A number of companies are currently developing
an infrastructure for trading bandwidth circuits which provides the
necessary measurement and routing capabilities.  At this writing,
however, physical trading of bandwidth circuits remains in its infancy
(see Figure 5. Pooling Points and Physical Matchmakers).

A physical matchmaker may deploy one or many switches or hubs and may
or may not interconnect them.  Clearly, more hubs extend an exchange's
geographic reach, allowing more buyers and sellers to interconnect
at a lesser cost.  Some exchanges have multiple, separate facilities
(e.g., Band-X); others are building a network of hubs that are linked
together (e.g., TheGTX).  Again, an exchange's network architecture
reflects the orientation of the service provided by the exchange, as
well as the exchange's preferred technological blueprint.  Even if
an exchange owns its own hubs, it may broker deals that are "off-hub,"
if it is more cost-efficient for the parties to interconnect through
their own facilities.

Owning facilities may allow a physical matchmaker to offer anonymous
trading services to its interconnected members and to measure actively
the traded minutes or circuits.  Monitoring of service quality,
in such instances, becomes critical for buyer confidence.  Physical
matchmakers may post quality metrics on their website or even
guarantee quality levels to buyers.  Band-X has pioneered its own
algorithm for determining the quality of IP network access that is
available through its facilities (again, see Figure 4).

A few physical matchmakers have developed facilities that allow
transactions to happen in "real time" -- instantly, without human
intervention.  Once connected to the exchange's hub, a member of
such an automated exchange can find a counterparty and provision the
service delivery entirely through a Web-based interface (see Figure
6. Case Study: Arbinet).

Payment and Risk

As exchanges differ by service and operational model, so too do
they vary by fee structure.  Buyers and sellers compensate bandwidth
exchanges through commissions on deals they facilitate, fees for
membership in the exchange, or revenue from advertising or other
ancillary services.

Consistent with practices in the energy industry, brokerage houses and
lead generators may generate commissions only from bandwidth sellers,
not buyers.  But not always.  Chapel Hill Broadband, for instance, has
a variable fee structure that depends on the type and volume of the
transaction it enables.  Chapel Hill's commission can range to as high
as seven percent, if assessed on both buyer (three percent) and seller
(four percent).

On its switched minutes trading floor, Band-X takes a five percent
total commission -- two percent from buyers and three percent from
sellers.  For "networks," or bandwidth circuit deals, Band-X charges
the successful bid/offer poster, whether buyer or seller, a two
and a half percent commission on the first $200,000 and one percent
thereafter.

Some exchanges have moved away from charging commissions altogether.
Arbinet, for example, has chosen to replace commissions on minutes
trades with flat-rate capacity-based fees.  Similarly, RateXchange has
forsaken commissions on bandwidth circuit deals in favor of increasing
its market presence.

Bandwidth exchanges can also take on counterparty risk in the
financial transaction.  In such cases, the exchange pays the seller
directly, regardless of whether it is able to collect payment from the
buyer.  The seller, therefore, is relieved of any worries of bad debt
or risk management.  Exchanges which offer these clearing services
"take title" to the minutes or bandwidth sold, becoming the buyer to
every seller, and the seller to every buyer.  For example, AIG Telecom
buys blocks of minutes from the selling carrier, pays the seller,
sells the minutes to the purchaser, and later invoices and collects
from the buyer.  To protect themselves, bandwidth exchanges often
require participating members to undergo extensive background credit
checks.

Summary

Bandwidth exchanges have a common goal: to facilitate the buying and
selling of communications capacity.  Most notably, exchanges help
companies deal in minutes, bandwidth circuits, and Internet access.

Exchanges provide virtual or physical matchmaking services -- or a
combination of the two.  Virtual matchmakers facilitate bandwidth
trades without physically linking buyers and sellers.  Of the
virtual matchmakers, bulletin-board operators serve as passive
lead generators, while OTC brokers play an active role in bringing
counterparties together.  Physical matchmakers, which have facilities
where their members interconnect, may offer anonymous trading and
monitoring capabilities at varying degrees of automation.

Payment structures of exchanges include differing commission schedules
and membership fees.  Bandwidth exchanges may assume counterparty risk
by guaranteeing payment to sellers.

Difficult as it may be to categorize bandwidth exchanges, the
industry's ongoing evolution is sure to bring greater standardization
as well as significant change.

TeleGeography, Inc. is a wholly-owned subsidiary of the bandwidth
exchange Band-X Ltd.  The editors of TeleGeography 2001 are solely
responsible for the accuracy and completeness of this article.

##

************************************************************************************
Distributed through Cyber-Society-Live [CSL]: CSL is a moderated discussion
list made up of people who are interested in the interdisciplinary academic
study of Cyber Society in all its manifestations.To join the list please visit:
http://www.jiscmail.ac.uk/lists/cyber-society-live.html
*************************************************************************************