The Oncoming Glut of Bandwidth
bcr.com
Volume 28, Number 8
August 1998, pp. 12-14
By John Puttr‚, (jputtre@aol.com), principal of Puttr‚, Inc., specializing in
strategic marketing, consulting and intelligence gathering for telephony,
computing, video and Internet companies.
The following is the full text of the printed article:
Only a short time ago, Dense Wave Division Multiplexing (DWDM) entered
the general investor's awareness when Ciena brought its IPO to market,
promising a product that would quadruple the bandwidth capacity of a single
fiber. On the day of its IPO, Ciena had less than $55 million in anticipated
revenues. When that first day of trading ended, Ciena had a market cap of $3.4
billion--60 times its entire revenue. It was the greatest one day's IPO in the
history of new technology, if not in the history of the stock market.
But that was just the beginning for DWDM. Most recently, Lucent
announced it could increase the capacity of a single fiber by 200 times. In the
interim, Ciena's market cap went to $7.1 billion as it combined with Tellabs. By
embracing this technology, carriers such as Qwest, Level 3, IXC and Williams
Communications are promising to change the economics of the telecom and
computer industry forever.
Bandwidth Availability
There are 49 million miles of fiber in North America, and until 1994, no one
seriously believed there was much need for more. It was assumed the fiber in
the ground could be modestly upgraded to handle any conceivable future
needs. Salomon Smith Barney's analysis of national bandwidth availability at
the time of AT&T's 1984 divestiture showed the monster telopoly had
balanced its available bandwidth capacity at 60 to 70 percent of all existing
bandwidth demand.
To build more might create a "glut" of bandwidth. What was the point of
investing money to build more capacity, which might only lower the resale
value of existing bandwidth? Demand was controlled by pricing set
unilaterally by the telopoly. If telecom users wanted more, then telopolies
could charge a premium, on the basis of scarcity.
After AT&T's breakup and the market's legitimization of MCI, Sprint, Wiltel
and others, the new carriers raced to build capacity to carry the traffic they
were capturing from AT&T. Though they built fiber backbones, their
technology was based on the same 64-kbps circuit-switched voice systems
AT&T had used for much of the century.
This was the original heyday of laying fiber without any regard to bandwidth
gluts. Protected by the AT&T pricing umbrella, the new telopolies had to
discount their rates only a few percentage points to get a lot of business. All
the new competitors' salespeople had to do was look up AT&T's tariffed rate
for any prospective customer and start cutting that rate at the start of their
sales call. It was like shooting fish in a barrel, and it was all so profitable that
new telopolies could be financed by the public and built overnight.
By the end of the new fiber-laying era, the combined bandwidth utilization of
all the telopolies fell to below 30 percent, despite an enormous increase in
usage, according to Salomon Smith Barney. Customer acquisition became
even more crucial, because carriers had to fill up these huge pipes, so the
telopolies turned to long distance resellers to get, steal or "slam" customers.
Hundreds of resellers sprang up overnight with little more than a
telemarketing office calling individuals and companies to sell heavily
discounted long distance services under new brand names, even though they
were only reselling the regular services of the telopolies. The real role of the
reseller was to find a buyer.
As a result, wholesale prices for bandwidth capacity as measured by cost per
DS0 for an equivalent mile per month on a one-year contract plunged from 22
cents a minute in 1988 to 4 cents a minute by 1995--a drop of 80 percent in
only six years, caused by the new competition and an increase in unused new
bandwidth of little more than 100 percent.
Yet by the middle '90s, the glut disappeared: The low cost of bandwidth and
the needs of the Internet once again created ever-greater demand for
bandwidth. So telopolies were able to raise bandwidth rates to business
customers; T1 prices increased 13 percent in 1997.
Now, because of DWDM and the intervention of the new carriers, bandwidth
is about to be added on an unprecedented scale. Nothing in
telecommunications, video and computers will ever be the same.
DWDM Problems for Older Telopolies
For older telopolies, the biggest economic problems are in legacy fiber
systems. Although more than 80 percent of the older telopolies' networks are
fiber, this plant varies in quality. Some may support only 1-Gbps speeds and
will not work with DWDM multipliers above four. Some fiber may not be able
to work economically at all with DWDM. What's more, there's a
lowest-common-denominator effect: contiguous fiber runs may vary in
quality, so when 10-Gbps fiber is connected to slower fiber, all speeds along
the network are reduced to the speed of the slowest fiber in the circuit.
Furthermore, to convert the older fiber systems to DWDM, the older
telopolies must use new electronic control techniques to manage the fiber.
Remember that wherever data enters or leaves the network, it's necessary to
convert the light beams into electrical signals. Therefore, upgrading to
DWDM requires physically getting to the old installed fiber to change the
electronics. Since the idea of DWDM did not exist when most of the legacy
fiber was laid, telopolies didn't anticipate the need to re-equip the fiber's
control systems, so they buried much of it without access points their
technicians could easily reach to upgrade old electronics.
Finally even if the older telopolies upgrade their fiber, they still may not be
economically competitive, because newer systems with purer fiber could have
many times their throughput. As such, the cost effectiveness of new
bandwidth vendors might be so great they could easily undercut the older
telopoly fiber networks--even after the latter have upgraded. In short,
upgrading their old fiber might just add to the "glut" and drive bandwidth
pricing even lower.
Qwest, IXC, Level3, Williams et al.
Telecom executives and financial backers outside the giant telopolies all
perceived the same symptoms and market needs. They saw a rapid tightening
of wholesale bandwidth markets and a constant call from the computer,
communications and video industries for more bandwidth at lower costs for
new applications. These telecom people understood the new economics of
DWDM bandwidth provisioning, along with the natural reluctance of old
telopolies to do away with their scarcity-based bandwidth thinking and
high-revenue business structures.
Initially, these new providers thought of themselves as super-bandwidth IP
carriers selling to the telopolies, who would lease the new bandwidth from
them when the old fiber backbones ran out of steam trying to handle the
traffic load of data growing at 1,000 percent a year (compared with voice traffic
growth of only 8 percent per year). But it quickly became obvious that the
newcomers' opportunities were really unlimited.
The projections were staggering. For example, Insight Research
(www.insight-corp.com) predicted worldwide bandwidth needs for data would
jump from 273 Gbps in 1997 to 27,645 Gbps by 2002--an increase of 101 times
in five years. Robertson Stephens projected this year that 25,000 T3s would
be needed by American industry in four years, compared with only 2,200 T3s
today. Meanwhile, cable networks came to the new providers, saying they
needed vast quantities of bandwidth for the services they hoped to deliver via
cable modems.
At the same time, more than 500 competitive local exchange carriers (CLECs)
had been licensed to compete with local access telopolies. Fifty CLECs have
raised almost $10 billion to build local access networks, according to
Robertson Stephens. Every one of these new CLECs needed to lease
wideband backbones to worldwide networks and to their customers. It's
unlikely any of them would consider renting their backbone from the very
local access or long distance telopoly they would be competing with.
Furthermore, there are 4,400 U.S.-based ISPs that are immediate prospects for
new wideband suppliers, because none of them can afford to be dependent on
a handful of telopolies that are or soon will be their competitors. These ISPs
all needed cheap wideband capacity to offer new value-added services.
The opening up of these new markets for bandwidth convinced the new
providers that they didn't have to limit themselves to being wholesale
suppliers of raw bandwidth to the old telopolies. Instead, they could compete
directly in selling services, or by selling bandwidth to others who were
undertaking to compete directly against the entrenched carriers.
New Economics and Implications of DWDM
DWDM theorists now believe a single fiber could transport 25 terabits per
second. Today, less than a single terabit would be enough to carry all of
North America's bandwidth needs. Using Qwest's new network as an example
of what's already being built, the economics and implications of DWDM
networks can be easily understood.
Qwest's network consists of two polyethylene conduits buried with carefully
placed access sites so technicians can easily upgrade the electronics
controlling the fiber. There is ample room alongside these conduits to lay
more conduits when needed.
Because the rights of way run along railroad routes, all construction is done
from specially designed railway cars using a "rail plow" to prepare the furrow
and then lower the fiber into its trench. Each conduit carries two cables with
96 strands each of OC-192-class fiber rated at 10 Gbps per strand when
running with only a single color (wavelength) or, as it is commonly called,
"one window."
When first designed, it was anticipated eight windows would run on each
strand, but the state of the art today is already 16 windows, so it should be
assumed Qwest could move up to 16 or perhaps even 40 windows eventually,
if market demand exists. To control initial costs and get immediate revenue,
Qwest has retained only 48 of the OC-192 fiber strands for itself and has
leased the other 48 strands to Frontier and GTE.
To get an idea of the bandwidth throughput of this single cable, multiply 96
strands times 10 Gbps per strand, times eight or 16 windows running. Then
consider that there is a second "dark" cable in the same conduit with the same
bandwidth capacity, and still another empty conduit space alongside that can
hold two more of these cables.
In fall 1997, Salomon, Smith Barney compared the Qwest setup--at a time when
it would reach just 17,000 route miles of fiber--to AT&Ts then-existing
network, in which the fastest fiber was only OC-48 (2.4 Gbps). AT&T has
41,000 route miles of fiber, of which it was estimated only 15 percent was of
the highest quality, and there are on average only 30 OC-48 fiber strands in
the higher-quality cables. Solomon Smith Barney therefore calculated that
even though AT&T would have 2.4 times more route miles of fiber networks,
"Qwest, if fully lit, in theory would have 20 times the throughput capacity of
AT&T."
But Qwest is not alone in feverishly laying fiber and raising billions of dollars
in public financial markets so they can keep on laying DWDM-based fiber.
Level3 recently raised $2.5 billion to add to its $3 billion "kitty" to keep laying
fiber, while Williams and IXC are also building similar networks. The Wall
Street Journal reported that these four new bandwidth vendors will have
63,000 route miles of DWDM cable laid by 2000. North River Ventures, a telco
industry venture-capital and consulting firm, has calculated that when all four
networks are complete, they will have 80 times the existing throughput
capacity of AT&T, when running only a minimal number of DWDM windows.
Meanwhile, AT&T is releasing press statements assuring that DWDM
technology will "make it possible for us to increase the transport capacity of
our existing network by a factor of 10, without having to lay any additional
fiber-optic cable." Sprint has promised that, "Through deployment of Dense
Wave Division Multiplexing and other fiber-optic technologies, Sprint [will]
efficiently and quickly scale network capacity.... In 1998, a single Sprint fiber
pair will be able to simultaneously carry over 2 million calls--the equivalent of
the combined peak time voice traffic of Sprint, AT&T and MCI.... In 2000, one
pair of Sprint fiber will have the capacity to handle 34 million calls
simultaneously." This indicates Sprint intends to upgrade its already fiber
backbone strands by a DWDM multiplier of 17 times.
Conclusion
The same dynamics that drove long distance prices down after the bandwidth
explosion of the previous decade are at work again--writ even larger. The
market forces that worked on voice in the 1980s will be brought to bear on
data and multimedia in the first decade of the 21st century. Retail bandwidth
pricing will have no place to go but down.
c1995-1998 BCR Enterprises, Inc. All Rights Reserved
This page was last modified on 08/10/98 Please direct comments, suggestions and problems to webmaster@bcr.com.
|
