Tom,
Interesting read from the Net Economy:
Glut of Bandwidth Analysis
By Carol Wilson
The notion of a bandwidth glut has been so widely reported that
it is now taken for granted. The reasoning offered by the Wall
Street Journal, The New York Times, Forbes and others is
straightforward: Too many companies have put too much fiber-
optic cable into the ground. Supply greatly exceeds demand,
therefore prices must fall.
The statistic most often trotted out for these editorial
exercises is one that may well be true: 98 percent of fiber laid
today remains unlit.
But as my teenager replies when I say his clothes are too baggy:
So?
DSL survivors hope to drive bandwidth demand:
newsletters.theneteconomy.com
I have long argued that this thinking overlooks the basics of
how fiber-optic networks are built. It is reasonable to bury
massive amounts of cable when building a network because the
costs of digging trenches, paying workers, moving equipment,
gaining rights of way and all the rest of what it takes to build
a network are the same for 800 fibers as for one. The only
incremental cost of the 800-fiber network is the price of the
fiber itself, and that remains a small percentage of overall
construction cost.
Here's Joe McGarvey's look at the way bandwidth use is affecting
metro networks:
newsletters.theneteconomy.com
But it is hard to continue pressing the point when the inches of
copy stack up against you. This past week, however, one of the
industry's leading consultancies, TeleChoice, weighed in on the
side of the "no-glut" forces with an in-depth study that shows
there is actually not even an abundance of bandwidth on many
inter-city routes, much less a glut.
The TeleChoice research was done on behalf of its client,
Williams Communications, says Russ McGuire, chief strategist at
TeleChoice and a former employee at Williams. The intent was to
go beyond anecdotal evidence of bandwidth supply to look at the
specific cities that Williams serves and the fiber routes that
link those cities to determine first, what the current supply is
and second, the different applications that are driving
bandwidth on those routes and how they are growing.
Want to know more about TeleChoice. Check out their site:
newsletters.theneteconomy.com
"The idea was to give Williams, first, a planning tool so they
could decide where to spend capital and second, a way to more
clearly communicate the value of their network to investors,"
says McGuire. Because TeleChoice wanted to use the model it
developed in its research for other clients, it agreed to pay
for half of the Williams project. Going forward, the consultancy
will also look at other possible facility "gluts" including
collocation space, hosting and data centers, international
gateways, metro networks and access networks.
Already, however, TeleChoice has turned up information that
directly contradicts the conventional wisdom of a bandwidth glut
within long-haul networks. When it examined the 22 fiber routes
that interconnect the 12 cities that Williams serves, it found
only three routes were truly overbuilt with fiber. On 14 routes,
capacity was already at 70 percent or higher, the point at which
telecom carriers traditionally start adding bandwidth to avoid
congestion or other problems.
Bandwidth demand may grow if broadband reaches the hinterlands:
newsletters.theneteconomy.com
The three routes that were overbuilt are three of the most
popular -- New York to Chicago, Denver to Los Angeles and Los
Angeles to San Francisco. They were overbuilt, McGuire says,
because every fiber-optic network operator had to have fiber in
those corridors. Even so, he says, that fiber will be will used
in coming years.
"This is somewhat a temporary problem," says McGuire. "Demand
does continue to grow. Even on the worst-case route we found,
which was Los Angeles to San Francisco, there will be plenty of
demand for this capacity."
Local traffic is still growing:
newsletters.theneteconomy.com
The highest demand is in Internet Protocol-based services, he
adds. IP is considered a more efficient means of transport than
other packet technology, such as Asynchronous Transfer Mode
switching, and certainly than circuit switching. But it still
has its overhead. One of the things that TeleChoice discovered
was that service providers use up a lot more bandwidth than just
applications consume.
"IP is very inefficient because of all of the extra stuff
flowing on IP networks -- BGP routing tables, IP signaling, all
the stuff in addition to what we think of as overhead, but which
is IP overhead," says McGuire. "An ISP has to provision a lot
more capacity than just the applications demand."
But if bandwidth is not in such oversupply, why are service
prices falling?
"There's probably two different factors," says McGuire. "Some of
what we hear is that the current market price on most of the
major routes is probably half anybody's cost, so everybody's
losing money on every route they sell. The issue is, how can
they keep doing that, is it because there is too little demand
for the supply, that's no. There is a shortage of services in
many places. If you want an OC-48 service, you can't buy it
because it is not available. But if you look at the price of
that service, it is ridiculously low."
What large network operators like Williams, Qwest
Communications, Global Crossing, Level 3 Communications and
others are trying to do by keeping prices low is to get their
network utilization up to a level at which they can begin to be
profitable, says McGuire.
"There are three factors at work here: technology, scale and
utilization," he says. Network operators have to use newer
technology that enables efficiency, they have to be able to
scale up their networks to drive down the cost per bit and they
have to get enough network traffic, or utilization, to begin
covering the fixed costs of network building and operation.
Even if they are losing money on getting that traffic, once they
reach a certain level of utilization, they can begin to sell
services profitably, McGuire adds.
John Charters talks his way to the top:
newsletters.theneteconomy.com
The question is who will be able to reach that level of
utilization and who will fall short. "It's a pretty steep
curve," he says. "Everybody is going after a few customers. The
CFOs have to be scratching their heads -- how can I make the
financial model work at this price?"
At the end of the day, the Journal and the Times are probably
right when they say that some of the current ambitious network
builders won't survive. Maybe it is mincing words to insist
their failure won't be the result of a bandwidth glut. But the
issues here are more complex than the current news coverage
allows, and I believe it is more important to see the complexity
than to reach the same conclusion based on different facts.
***************************************************************** |
