Optical Delusions:

Posted: 2/2003

^{Optical Delusions:}
Wavelengths Slow to Reach Potential

By Khali Henderson

YOU KNOW THOSE FIREWORKS that shoot
way up in the sky, explode with a bang and then literally fizzle out? They are a
little like wavelength services. They had a bright future when they burst on the
scene four years ago — analysts and vendors alike predicted they would be in
great demand in the long haul and the metro area, from carriers and large end
users alike. Today, the sparkle is fading into the darkness of telecom’s
downturn, but industry experts are not ready to say the show is over. They
predict waves to light up more brilliantly over the long term.

While wavelength services have
enormous potential applications–to extend carrier networks economically and to
facilitate on-demand bandwidth for enterprise video and data storage
applications–their use remains limited almost totally to point-to-point
unprotected wavelengths at 2.5gbps (OC48 equivalent) and 10gbps (OC192
equivalent).

That wavelengths have not lived up
to their early promise can be attributed to overcapacity and the telecom slump
that has slowed metro DWDM deployments and vendor Research and Development.

"There is plenty of demand [for
capacity], it’s just not as great as the supply," says analyst Paul Kellet,
senior director of research for Pioneer Consulting LLC, who notes the telecom
sector will recover at the point at which demand and supply intersect. When will
that be? Kellet projects recovery will begin in fourth quarter 2003 with
positive growth in 2004.

Even when over-capacity issues are
factored in, Kellet sees a longer-term market opportunity for wavelength
services, and projects revenue to reach nearly $1.6 billion by 2007 (see chart,
"Global Wavelength Service Revenue Forecast," above). This growth, he
says, is based only on static, unmanaged services like those that are sold
today. He explains second-generation dynamic, or switched, wavelength services
are another four or five years away, held up primarily by a lack of standard
protocols and a sound business case.

Alternatives to traditional optical
add/drop multiplexers are emerging that may speed up the viability of switching
wavelengths and the deployment of optical virtual private networks (VPNs).

"Almost all wavelength services
have been route-by-route swaps between carriers," says David Gross, a
senior analyst for Communications Industry Researchers Inc. (CIR)
"Ultimately, wavelengths will need retail demand to support its
growth."

Retail calls for metro services.
Gross says there is not enough DWDM, which is required to provision wavelength
services, deployed in metro areas. CIR’s research shows 2003 will be the first
big year for substantial metro DWDM deployment to occur in the United States,
with the market reaching $500 million in 2005. This growth hinges on the Bells,
because many optical Ethernet startups, like Yipes Enterprise Services Inc. and
Telseon Inc., which emerged to provide an alternative to SONET-based ILEC
networks, have gone out of business.

One of the few remaining metro
optical competitors, OnFiber Communications Inc., which acquired the assets of
both Telseon and Sphera Optical Networks Inc. in 2002, reports retail demand
from content providers linking data centers to create private networks within a
metro, says Boyd Chastant, product manager for OnFiber. He says OnFiber has
helped a carrier customer extend Fibre Channel for an enterprise storage area
network (SAN) application, but that SANs have not drawn the interest that
initially was projected. He predicts the recent passage of the Homeland Security
Act might drive additional Fibre Channel connections among large enterprises,
such as financial institutions.

With the exception of extending the
corporate LAN with gigabit Ethernet (GigE), Memphis Networx LLC is not quoting
wholesale wavelength services in support of bursty enterprise applications like
SANs or video streaming, or video conferencing either. Charles Elliott, director
of sales and marketing for Memphis Networx, says most enterprise apps are still
being quoted over SONET.

"There’s no out-of-the-box
thinking because of capex shrinkage," Elliot says, saying most carriers are
doing what they can just to keep the customers they have.

Analyst Gross says it’s a matter of
economics. If an application requires an access pipe, OC3 to OC12 is typically
sufficient. "It makes no sense to offer a wave below OC48," he says,
noting that enterprise requirements for that bandwidth are limited.

Meanwhile, OnFiber and Memphis
Networx are wholesaling wavelengths primarily to IXCs seeking to extend their
networks into the metro core and metro access networks.

"It’s a way for IXCs to extend
their networks without capex," says Chastant, who says OnFiber gets half
its business from selling wavelength services. "The dilemma is if they
lease the network to get the customers, when do they build out?"

He says many of OnFiber’s carrier
customers are struggling with this question. He tells them that pulling and
lighting dark fiber so it can depreciate it over three to five years does not
take into account the total cost of ownership, including metro equipment and
operations, as well as the expertise to manage both. "We find it’s not so
simple to move around the metro, coordinating with multiple carriers and,
potentially, buildings," Chastant says, explaining that IXCs’ focus usually
is back into the long-haul network and not into the metro.

Elliott says wholesale wavelengths
remain less than 20 percent of the Memphis Networx’ business, but about half are
GigE and half are for IXCs connecting from central office to central office in
the metro. "Normally they would use BellSouth [Corp.] for that, but now
that it has 271 relief [to provide in-region long-distance services], they are
looking to us as a neutral party," he says. Elliott predicts this part of
Memphis Networx’ business will grow in 2003-2004, now that BellSouth has gained
271 approvals in its entire territory.

Global Wavelength Service Revenue Forecast
Source: Pioneer Consulting LLC