Just a matter of time before they put cellular systems in all the subway systems.
Another interesting read.
zdnet.com
Building A Better Backbone - And Business
Plan
By Carol Wilson, Inter@ctive Week
September 6, 1999 9:23 AM ET
The numbers are staggering: Six major new national fiber-optic
networks will come on line during the next two years at a cost of $18
billion. Those new networks are in addition to those already built and
operated by the Big Three of long-distance: AT&T, MCI WorldCom
and Sprint.
Thousands of miles of fiber-optic cable are being installed in ducts to
which new fiber can be easily added by companies such as Enron
Communications, Frontier Communications, IXC Communications,
Level 3 Communications, Qwest Communications International and
Williams Communications.
And that's just a piece of the story.
While competition in the backbone network business is booming, new
technologies coming into the market will make it much easier to
increase bandwidth available on any fiber-optic network tenfold or
more. Dense Wavelength Division Multiplexing (DWDM) systems now
exist to put data traffic onto 160 different wavelengths on a single
fiber-optic cable. Technology exists today to enable each of those
wavelengths to transport 10 gigabits per second of data. Within two
years, that will reach 40 gigabits per wavelength per second.
"In effect, the backbone of the Internet is being rebuilt. The next step:
Recover all those costs of construction,'' says analyst Peter Bernstein
at Infonautics Consulting. "And that step may be more difficult than
building the networks themselves.''
The combination of major new network construction and new
technology for adding network capacity has led industry analysts and
executives to predict a bandwidth glut that is likely to drive down
prices in backbone transport of data much as the price of
long-distance telephony has plummeted in the past few years.
Frontier Communications admitted as much earlier this year when its
second-quarter earnings weren't up to expectations.
Rather than a free fall in prices, however, expect the market to
quickly find new ways to use bandwidth, both in terms of
network-based applications and bigger backbone networks to support
the rapidly growing field of faster access via Digital Subscriber Line
(DSL) and cable modems. The challenge for network operators is to
create new business models that capitalize on something other than
cheaper bandwidth.
"I think the danger isn't so much of a glut as a lack of planning for the
backbone network that could cause some major problems," says
Claudia Bacco, an analyst at TeleChoice who studies the
development of DSL. "Once DSL starts selling as a volume service, it
will very quickly increase the demand for bandwidth in the backbone.
Service providers need to have planned ahead."
It's not surprising, then, to see a growing number of alliances between
the new backbone builders and the competitive local exchange
carriers (CLECs) that have been building DSL networks in major
markets across the U.S.
Qwest invested in both Covad Communications and Rhythms
NetConnect-ions, for instance, and a third data CLEC, NorthPoint
Communications, includes Enron, Frontier and Level 3 among its
strategic partners.
That doesn't mean, however, that all of the companies investing
billions in burying fiber-optic cable stand to reap rich rewards,
cautions Bernstein at Infonautics.
"The problem isn't a glut of bandwidth, but a glut of service
providers," he says. "If everyone has the same business model and is
going after the same customers, that's a problem. In addition, there
are timing issues . . . If you build it, they will come - but not
necessarily when you need them to."
At least nine players have built or are building national fiber-optic
backbone networks. These look very similar, says Fred Harris, director
of network planning and design at Sprint. "There are generally three
East-West routes and three or four North-South routes that everyone
builds," he says. "New York to San Francisco, Atlanta to Los Angeles,
Seattle to Boston - everybody has fiber either in or near those
corridors."
In earlier network deployments, the fiber was buried, never to be
exhumed. But today's fiber cables go into ducts, allowing fibers to be
pulled out or slid in later. Capacity can be added without any
backhoes.
Today's newer fiber-optic cable, from companies such as Corning
and Lucent Technologies, is designed to support higher transmission
speeds without the light dispersion problems that limited speeds or
distances on other fiber-optic cables.
DWDM technology and optical amplifiers, which boost the power of a
laser's signal over a given wavelength, enable bandwidth to be added
and allow signals to travel farther without the need for as many
repeaters to boost signals along the way. Eliminating repeaters in a
long-distance network wrings out installation, maintenance and power
costs.
As Harris points out, if every company building a national backbone
has 30 fibers along each major route, and each fiber can carry 80
different wavelengths of light, that's more than 20,000 different
signals. Today, most of those signals would be powered by
Synchronous Optical Network (SONET) gear, which operates at 2.5
gigabits per second, but a growing number have moved up to gear
that operates at 10 Gbps.
Both Lucent and Nortel Networks have announced equipment that
supports 160 wavelengths on a single cable, and other vendors will
have similar equipment within the next two years.
Also coming into the market by 2002 will be new SONET gear that
quadruples the current transmission rates from 10 Gbps to 40 Gbps.
MCI WorldCom is testing these systems, known as OC-768 transmission
systems, and expects to have one commercially available by early
2001, says Rama Nune, senior manager of optical and data network
technology development.
Nune stresses, however, that what's more important than having all
the bandwidth is knowing "how to intelligently provision services over
it and manage it."
New Models
Not surprisingly, most of the newer network operators have quickly
moved to talking about services, not bandwidth.
There are exceptions to this: Enron is building a network that
transmits Internet Protocol (IP)-based traffic directly over a fiber-optic
network, with no intervening transport systems or protocols. But its
network is built to provide backbone bandwidth for Internet service
providers and content providers, says Stan Hanks, vice president of
research and technology at the company.
At Level 3, falling backbone service prices would be good news, says
spokesman David Powers. "We want to drive prices down by reducing
the underlying cost of the network," he says. "We think we can do it
because we are building a green-field network that doesn't have the
cost of legacy technology like circuit switches." Most other network
operators are looking to add services to their mix, including Web
hosting, to create a differentiated service. MCI WorldCom through its
UUnet subsidiary, Frontier, Qwest and others have announced Web
hosting services.
Greater bandwidth and lower prices also will enable new video
services to be delivered over the Internet, says Vab Goel, vice
president of IP engineering and technology at Qwest.
"One of the things Qwest is doing is building its own access rings in
25 major markets," Goel says. "We think this network will drive cost
down and increase performance for value-added applications which
can be delivered over an IP network. Today, most video is
compressed. With faster access speeds, you can watch TV shows over
the Internet. This becomes interesting when you talk about having
access to any TV show broadcast anywhere around the globe."
In addition to enabling new services, however, the sudden availability
of bandwidth will dramatically change the way services are priced
and sold.
"It's the next great challenge of the Internet - deciding how you are
going to bill customers and what they will be willing to pay for,"
analyst Bernstein says. "I'm not sure anyone knows yet."
Part of that challenge is matching a service that customers value with
the need to recover the high costs of expanding a network.
"George Gilder likes to say that bandwidth will be free - well, putting
in 20,000 channels of OC-48 [2.5-Gbps transmision] isn't free; it costs
billions," Sprint's Harris says. "And the same thing goes when you
build up the local copper network to do DSL or cable modems."
Already, however, DSL prices are dropping to $30 to $40 per month
for a consumer service that offers up to 1.5 megabits per second.
Businesses still typically pay $125 to $400 per month for higher-speed
services that offer the same bandwidth in both upload and download
modes. That compares very favorably to the $2,000 per month most
companies paid for a T1 line until the past two years.
But with customers assuming bandwidth will be cheap, how will nine
backbone network providers recover their billions in investment?
"I don't have a crystal ball," says Nune at MCI. "With the access
speeds going up - whether it be ADSL [Asymmetric DSL] or gigabit
Ethernet - we anticipate many rich applications to drive the
bandwidth of our customers."
Providing those applications - or at least building into the network the
right hooks for new services - is the best hope for building a new
revenue base, Harris agrees.
There will still be room for charging higher rates as well for levels of
reliability and transmission speeds, says Qwest's Goel, but those
charges will likely be packaged as part of the application cost.
In a world where everyone has access to the same technology, it will
be the business issues that ultimately decide who succeeds at
building - and sustaining - the backbone of the next Internet. |
