Local Networks Grow Fiber as
Bandwidth Demand Escalates
By Ken Branson
In The Graduate, a stiff, middle-aged, corporate suit
pulls the newly graduated hero aside at a party and says, with
considerable buildup, that he has one word for the hero: "Plastics,"
he says, in a tone suggesting that he is passing on the secret of
life itself. Today, he might have two words for the graduate: "fiber
These are flush times for anybody who does anything with optical
fiber--making it, cabling it, providing the electronics or
"optronics" that make it go, building the networks, or leasing and
selling the fiber in those networks. If you work for a CLEC
installing metropolitan networks, you know this because you have to
make lots of decisions revolving around fiber: Should you lay it
yourself or hire someone else to lay it? Should you bury it or hang
it from towers? How seriously should you take optical networking?
How close is an all-optical switching environment? Does it matter
what kind of fiber you buy? Can you ever have too much bandwidth?
There seems to be a consensus on the answer to the last question:
"Emphatically no," says Charlotte Denenberg, vice president of
optical networks for Metromedia Fiber Network Inc. (http://www.mmfn.com/). "I think
we're at the beginning of a curve that looks more like a 90-degree
angle than a curve. We've demonstrated in previous paradigm shifts
that things start by people just doing the same old things better.
... That's where we are now. The real test comes when you're doing
new things. Bandwidth demand is nothing compared to what's coming."
MFN is building local dark fiber networks in 67 cities around the
world, including 51 in the United States, of which 17 are currently
operational. The company is in the business of selling metropolitan
dark fiber to other carriers, government agencies and large
corporations, and recently passed the $2 billion mark in lease
agreements. In the process of all this activity, MFN, its customers
and competitors are also redefining the word "metropolitan."
"We build fiber optic networks where it's hard to build them--in
metropolitan areas," Denenberg says. "But we follow our customers to
the suburbs, or up and down corridors. So we view New York to
Washington as a major metropolitan network, not as a corridor."
The question of whether to lay one's own fiber or hire someone
else to do it has to do with how much money you have, and how much
time. But nobody has time; by the time you notice a market window,
at least one competitor is preparing to leap through it.
So let's take a look at money.
Build, Hire or Buy?
Building a metropolitan fiber optic network is expensive. Optical
fiber itself doesn't come cheap, and neither do the electronic and
optronic gadgets to make it go. But the real expense, according to
Mike Miller, is constructing it. Miller is the president of ACSI
Network Technologies Inc. (http://www.acsint.net/), a
subsidiary of e.spire Communications Inc. (http://www.espire.net/). ACSI is
responsible for building fiber optic networks for all the major
CLECs, long-distance providers and Internet backbone providers
around the country today.
Miller's case for hiring ACSI or a similar company is that most
CLECs don't have engineering and construction people on their staffs
and can't afford to hire them. He concedes that a CLEC might just
hire a construction firm and tell it to build a network, but not
surprisingly, he thinks that would be an expensive mistake.
"We're a communications company that provides construction; not a
construction company building networks," Miller says. "Either one
could build the network. The difference is, I understand what the
pieces are for."
Miller wouldn't quote a price range for building a metropolitan
fiber optic network. Local conditions vary too much, he explains.
But he does say his customers have a choice of being charged a fixed
price for the entire optical project, or being charged on a unit
basis. By "unit basis" he means his customer sees a bill with each
expense broken out: labor, materials, permits, engineering,
"as-built" drawings, and a project management fee of between 15
percent and 20 percent. This allows the customer a little management
control over the project and its expense, but it requires a customer
who knows enough about construction to understand what he's
For Denenberg, of course, the whole idea of hiring someone to
construct a fiber optic network is crazy if one already exists in
the city a CLEC wants to enter. "I have not encountered an argument
for building your own in places where someone else is already
supplying," she says. Construction takes a lot of focus, and CLECs
are focused elsewhere, she says. "If they become encumbered with
things that are not central, they become ILECs," Denenberg says.
Miller says he has encountered such an argument, more than
once. ACSI, besides building networks for other people, builds some
for itself to sell as dark fiber. He says customers have asked him
to build brand new fiber optic networks in cities where he has
already built one.
"The most common error that I see ... for the CLEC industry is
... [not] taking advantage of the infrastructure that is in place
today that provides them the ability to open up new cities quickly,"
Miller says. "Instead, I see business plans that require large
construction processes that delay their entering the market ahead of
their competition. They get a business plan, decide they're going to
spend $200 million and then decide to construct in a city that
already has infrastructure in place. The 10 or 12 CLECs that are in
Atlanta today couldn't use up all the fiber that's already there
over the next 20 years."
Such decisions may result from too much venture capital money
chasing too few brains. But Miller points out that some carriers are
"trying to build asset value"--that is, they want to carry the new
network as an asset on their books, not as an expense, because that
makes their bottom lines look better.
Denenberg points out that a CLEC can buy dark fiber from MFN or
one of its competitors and solve that problem. When Denenberg and
her colleagues talk about "selling" dark fiber, they're really
talking about selling an indefeasible right of use (IRU). Under an
IRU, a carrier pays MFN a certain amount of money for the privilege
of controlling a set amount of fiber in a particular place for a
long time, usually 15 to 25 years. The carrier can account for the
fiber as an asset rather than as an expense.
Andrew Walker, president and CEO of ITC^DeltaCom Inc. (http://www.itcdeltacom.com/),
says his company has deployed fiber in just about every possible way
over the past few years.
"We've leased dark fiber," he says. "We've purchased dark fiber.
We have constructed. We've done collaboration with utility
companies. ... They built for us on their right of way. Rather than
capitalizing, getting a crew, paying a right-of-way provider, we
contacted the power company to place aerial fiber."
ITC^DeltaCom, which also operates a regional long-haul network in
the Southeast, prefers aerial fiber over buried fiber, Walker says.
A carrier that buries its fiber using backhoes and construction
crews can count its daily progress in yards, he says. Aerial fiber
is deployed by helicopters, and the carrier that deploys aerial
fiber can count its daily progress in miles. And aerial fiber,
specially designed and constructed for such deployment, is as secure
as buried fiber. Backhoes, for example, almost never climb towers.
Walker claims his aerial fiber has stood up to worse than backhoes.
"Two years ago we had a tornado in Alabama that took out two
towers," he says. "They cut the power off, but light kept going down
our cable. Now, if a 747 hits the tower, we've got a problem."
ITC^DeltaCom recently finished building a corridor--what
Denenberg and her colleagues would call a major metropolitan
network--between New Orleans and Tallahassee, Fla. Walker says there
are four segments in the network, and almost a different method for
building each one: a fiber swap with another carrier, an IRU from a
power company, a stretch of ITC^DeltaCom construction, and a section
of collaborative construction with another carrier. "That's typical
throughout the network," Walker says.
Is the Glass Getting Better?
Optical fiber itself now comes in a number of different
varieties. Companies that make fiber, like Corning Inc. (http://www.corning.com/), usually
sell fiber to cabling companies, who put it in sheaths in densities
ranging from 12 fibers to hundreds. But the fact that they sell to
cabling companies doesn't mean they don't market to carriers.
"We sell our fiber to people who cable it, but we talk a lot with
end users because they're the people who understand how it works,"
says Jane Li, marketing and sales director for metropolitan products
"There's an explosion of [fiber] types," Denenberg says. "Lucent
[Technologies Inc., http://www.lucent.com/] with
Truewave, Corning with MetroCore. There's a lot of hoopla, so
customers start to ask for new varieties, but there's no
[optoelectronic] equipment for it. So will you go with the newest
fiber, or one that fits the equipment profile?"
Of course, Li contends that MetroCore is the ideal product for a
metropolitan fiber network. Up to now, single-mode fiber has been
designed to operate at a wavelength of 1,310 nanometers, which makes
it harder to use DWDM, which operates best at 1,550 nanometers, she
says. But MetroCore is designed to work at 1,550 nanometers. She
insists there are optoelectronic products that work at that
Li would not say how much MetroCore costs, or even discuss the
basis by which Corning charges customers for it. Walker says his
fiber suppliers charge ITC^DeltaCom by the foot.
Devices: Is the All-Optical World Here?
ITC^DeltaCom's network, like most telecom networks, is a
many-layered thing. It's a SONET ring network, with ATM, frame relay
and IP added. Walker would be happy to reduce the complexity a bit.
All photonic switching promises to do that, but Walker is cautious
about when that will happen for ITC^DeltaCom.
"Photonic switching will come. It will be viable, but it won't be
the end-all," he says. "But the key is, the world of SONET is [a]
very secure, safe and high-quality service world. We dwell there,
not because we don't understand optical switching, but because we
think its time hasn't come."
There are others who disagree, though not always from
disinterested motives. They point out that the universal hunger for
bandwidth is straining public network resources, whether operated by
telecom companies or cable companies, and that the strain will only
Doug Sawyer, product line manager at Artel Video Systems Inc. (http://www.artel.com/), says video
on demand, many times born and many times buried in the past 15
years, has sprung to life again, thanks to video compression and
DWDM. Telecom carriers, he says, will soon use their infrastructures
to deliver compressed video content, and that content can be managed
by an MPEG switching platform--which Artel just happens to have. The
platform is CrossStream, and Sawyer says it's "a piece of equipment
that knows what MPEG video is and is able to switch that onto a
public network, ATM or IP."
Tim Dixon and Luc Ceuppens argue that all the necessary elements
already exist for all-optical networks; it's just a matter of
building them. Dixon is vice president of marketing and Ceuppens is
senior director of product management for Chromisys Inc. (http://www.chromisys.com/), a
Sunnyvale, Calif., startup whose first product, the Diamond Wave
photonic switch, was scheduled to be ready for market this spring.
They argue that a photonic switch, by eliminating the layers of
complexity found in current telecom networks, might make the
transport infrastructure--the fiber--easier to own than to lease or
IRU. Their argument boils down to this: Photonic switches are the
natural engine for fiber optic networks, and fiber driven by digital
switching is a little like early steamships using wind power for
most of their propulsion.
Denenberg says MFN isn't looking at optical switching just now.
She sees issues of scalability and reparability. And then there's
the big one for a network operator with QoS guarantees to consider:
"How do you monitor this stuff?" she asks.
Dixon contends that he and his colleagues are not unacquainted
with the real world.
"We're not photon bigots," he says. "We realize there is an
infrastructure already built out, from voice 5ESS and SONET, but
what we see is migration over time. We design our switch to help
bridge from digital, SONET-oriented infrastructure to all-photonic
infrastructure over time. We do that by letting them [carriers]
selectively install I/O [input/output] cards that support SONET and
all the performance monitoring, and then let them migrate to an all
photonic set of cards, as appropriate, in their network."
"Digital switches have value, depending on the services,"
Ceuppens adds. "If you're offering a wide array of services, you'll
need those digital switches to break down and groom back all those
lower speed services into higher speeds. Now, on the other hand,
there is enormous pressure to offer high-speed services at OC-48 and
above, and the digital equipment is really struggling. Let the
photonic switches handle the high-speed connections, let the digital
handle the lower and connect the two with an appropriate I/O card."
Ceuppens and Dixon concede, however, that they are not going
after carriers with large digital infrastructures. Regional CLECs
like ITC^DeltaCom might be target customers, they say, but their
real targets are "new generation CLECs that are focused on broadband
services, on bringing 100mbps pipes or bigger to the end user."
But for now, Denenberg has her eyes on those 67 metropolitan
fiber networks she and her colleagues are publicly committed to
"The telecom industry always has the next solution just
when you're getting traction with the current solution," she
says. "It's always just a little stuff around the edges, but when
you're trying to serve lots of customers in a high-quality,
responsive way, it's those edges that make the difference."