|
From the article "Into the Light" By David Sandham of IET Knowledge Network; Copyright, April 7, 2009.
In the second of two features on the history and future of the fibre market, E&T asks whether we are on the brink of a new fibre-optic
network revolution.
The telecommunications industry built too much fibre-optic network capacity during the 1990s. Companies such as KPNQwest, FLAG,
Global Crossing and Tyco borrowed too much money and laid too much fibre, and suffered the consequences. Is that over-capacity
still present? It’s not clear.
Most would agree with Jeff Ferry of optical network company Infinera when he says that “there is spare capacity in optical fibre
between most major cities”.
But how much? Mike Stein, CEO of Darkstrand, a network provider, says that about 80 to 90 per cent of the world’s fibre is unlit, that
is not being used to carry data. Julian Rawle, managing partner of Pioneer Consulting, says that only 18 per cent of world submarine
fibre capacity is lit.
Prices for fibre-optic network capacity collapsed when the Internet bubble burst. Stein says they have remained relatively flat since,
though there “has been a spike” since about a year ago.
Vinay Rathore, senior director of marketing for EMEA at optical networking equipment maker Ciena, disagrees that overcapacity is a
pressing issue.
“I don’t know anyone who believes overcapacity is a looming problem,” he says. “The bigger concern is that we are not increasing
capacity fast enough, while at the same time reducing cost per bit. There are places where dark fibre is plentiful, but also many where
it is not available at all. Radically different prices reflect this reality.”
Part of the difficulty with knowing the true use of fibre-optic networks is that the price of the bandwidth they offer, which could be used
as a proxy for use, is not freely available. Consultancy TeleGeography is one of the few sources of information on fibre pricing.
“No one really knows for sure what people are paying for their bandwidth because carriers are extremely careful about giving out that
kind of information,” says Rawle.
Precious dark fibre
Stein suggests that telcos are sitting on masses of dark fibre, which they do not want to release. But Rawle disagrees with this view,
in which the telcos are like oil barons, carefully managing the supply of a precious resource. Telcos want to make sales, but don’t
want to lease dark fibre to companies that would sell the capacity on, because they would prefer to sell a high-value service than a
commodity infrastructure.
Chicago-based Darkstrand gets around the telcos’ reluctance to sell dark-fibre capacity by sourcing it from a research network
instead. National LambdaRail leases dark fibre and uses it for supercomputing applications that demand huge amounts of bandwidth.
But National LambdaRail does not need all of its capacity at the moment and so has leased 80 lightpaths of 10Gbit/s each to DarkStrand.
Darkstrand has since talked with companies, including Caterpillar, State Farm, Disney and Sony, about selling them this capacity.
According to Stein, who founded Darkstrand, he can offer capacity more cheaply than telcos: “Telcos are not crazy about our model,
but we are not competing with them for residential customers. Major corporations are building their own dark-fibre networks. If they
were being serviced properly, they would not be building these networks.”
Others disagree about the advantage to a business or government of owning dark fibre. Stu Elby, vice president of network
architecture for Verizon’s technology group, says: “The cost of optical transmission is almost entirely in the electronics and operations,
administration and maintenance of these electronics, not the fibre itself. The scale of a large carrier could always bring a lower total
cost of ownership to a fibre route than a smaller government or business could. There are exceptions – special speeds or coding
schemes that certain government agencies may require that cannot be purchased from a carrier.”
Rathore emphasises the importance of scale: “If a business or government needs large amounts of bandwidth and they have access
to dark fibre at a reasonable cost, as well as the capability and desire to manage their network infrastructure, building their own
networks on dark fibre will provide the ability to scale capacity at very low incremental cost.”
National LambdaRail does not own the dark-fibre capacity it sold to Darkstrand, but leased it from Level 3 in a 20-year deal that has
some 15 years left to run. Some other research networks, such as Internet2, are not allowed to resell unused capacity, and Stein says
there are rumours that AT&T is pressing research groups not to commercialise their dark fibre.
Internet growth
The Internet was born out of many academic and research networks, and Stein believes that a new superfast Internet is emerging in
much the same way today. Video, file sharing, cloud computing and other high-bandwidth applications are driving demand for a
second-generation Internet.
This is very different from a decade ago. As Rathore says: “In 1998, we were deploying 16 or 40 [channel], 2.5Gbit/s channel
systems. Now channel rates are 40Gbit/s and going to 100Gbit/s.”
Rathore believes that long-term growth in bandwidth demand has been running at about 50 per cent a year. Others support that
estimate. AT&T carries more than 15.8 petabytes of IP and data traffic on an average business day, and says that its traffic volume is
growing more than 50 per cent a year.
For Stein, it is significant that President Obama has included Eric Schmidt, the CEO of Google, among his advisers. “The digital
society that everyone predicted [in the late 1990s] is now here,” Stein says. “It just happened more slowly.”
Is the current economic downturn affecting demand for bandwidth? Elby says not: “We have not noticed any bandwidth demand
slow-down.”
Rathore agrees: “The fundamental drivers of our customers’ business remain strong.”
New technologies
Technological advances have increased the capabilities of existing fibre.
“
Customers are demanding speed and flexibility in optics that were not commercially feasible ten years ago,” says Rathore of Ciena.
In response, new technologies have evolved. One of the most important is the reconfigurable optical add-drop multiplexer (ROADM),
which enables automated, all-optical wavelength switching between different fibres. In wavelength-division multiplexing (WDM) many
optical carrier signals are carried (or multiplexed) on a single fibre by using a different wavelength of light for each.
AROADM can add or drop traffic to or from a fibre entirely in the optical domain, which is simpler than converting the optical signals
back into electronic signals. For Rathore, ROADM is one of the most important technologies of recent years, as it “evolves the WDM
layer from simple point-to-point capacity delivery to an automated and flexible network-switching layer”.
Rathore says that running the Ethernet networking protocol over optical networks is also likely to become important.
“Ethernet over optics seems to be the leading contender for the long term,” he says.
Ethernet over optics involves combining Ethernet framing with optical networking. Almost all corporate Intranet traffic uses Ethernet in
the local area network (LAN). As the trend to share information beyond the LAN grows, it makes more sense to use Ethernet on the
optical network, removing the boundary between the LAN and the wide area network.
“Cloud computing and storage virtualisation will be key drivers pushing data that typically ran on a LAN to the wide area network,”
says Rathore.
Elby agrees about the importance of Ethernet over optics: “The industry is on the verge of converging Ethernet and packet capabilities
into optical platforms. This convergence has significantly reduced operating and capital expenses for the carriers that take advantage
of these advances.”
Ferry of Infinera believes that technological development is more significant than fibre build. “Fibre is not the major constraint to
expanding network capacity economically,” he says.
“Each optical fibre can carry tens of terabits per second of capacity and there is spare capacity. The constraint in many places is due
to insufficient capacity in the optical systems, or other networking systems.
“Our breakthrough in photonic integration [an ROADM] is important because it helps to make capacity additions to the optical network
easier and more cost-effective.”
Enough is enough?
Åke Sundin, technical manager in the network department of AB Stokab, which manages Stockholm’s fibre-optic network, disagrees
that the world has enough fibre.
“Fibre-to-the-home projects have just started,” he says. AB Stokab’s Stokholm network is made up of 5,600km of cable carrying
1,200,000km of fibre. Sundin believes that the key developments today are in the access network. “Building the access network will
put more demands on the metro or city networks that currently exist either to increase fibre counts in cables or to implement
cost-effective electronics,” he says. Stockab will be increasing fibre counts.
Elby of Verizon agrees about the importance of the access network. “Fibre to the premises is driving new fibre construction in the local
networks.” He says that Verizon has very little spare long-haul fibre, with most of it being lit using 10Gbit/s technology on fewer than
80 wavelengths per fibre. “We are upgrading these to 40Gbit/s on 128-wavelength systems, so there is plenty of growth capacity.
Also, we anticipate upgrading to 100Gbit/s beginning in 2010.”
Submarine revival
Submarine fibre-optic cables are being laid again, especially in the Asia Pacific, South Asia, India, Middle East and African markets.
For example, the Trans-Pacific Express (TPE), a $500m fibre-optic submarine cable system, was launched in September 2008.
TPE
connects the United States, China, Korea, Taiwan and Hong Kong over a total of 18,000km. The new link will increase the capacity
between the US and China sixty-fold. It is designed to carry up to 5.12Tbit/s, though it will be configured to handle 1.2Tbit/s at first.
Customers can buy 10Gbit/s links.
TPE is the first fibre-optic submarine cable to land on the west coast of the US in six years, since the Tyco Global Network (TGN) was
commissioned in 2002.
Other new links include Asia American Gateway, a 1.92Tbit/s fibre-optic system due to enter service soon, connecting the US,
Malaysia, Singapore, Vietnam, Hong Kong and the Philippines.
Europe India Gateway is a 3.84Tbit/s submarine fibre-optic cable project, due to be completed by the second quarter of 2010, which
will land in the UK, Portugal, Gibraltar, Morocco, Monaco, France, Libya, Egypt, Saudi Arabia, Djibouti, Oman, United Arab Emirates
and India.
Google has combined with other companies to build a trans-Pacific fibre-optic cable called Unity. According to Rawle, Google had the
idea because it is building a server farm in Malaysia which it needs to connect to the US. “They went into the market but could not get
the price they wanted so came up with the idea of a mini-consortium,” he says. Unity will be built by Tyco and NEC for $300m and will
provide 4.8Tbit/s.
More than 25 submarine cable systems are due to be built over the next three years. According to consultants T Soja, over the next
few years 85,000km of cable will be installed annually, approaching the 100,000km that was installed in 2001.
This revival, if it survives the economic downturn, may benefit from learning the lessons of the first Internet boom and bust. There are
signs that it has. In the late 1990s, for example, suppliers rushed to fill demand as quickly as possible. Nowadays, says Rawle, “in the
face of rising demand, Alcatel and Tyco [the two leading submarine cable suppliers] are pushing out delivery dates to 18 months or
even two years, which is unheard of”.
This dampening effect could be beneficial: “This is the only thing that will save the submarine [fibre] industry from another big bust at
the end of the cycle.”
The opportunities for fibre are greatest in the Asia Pacific region where strong demand is mismatched with low penetration. The oil-rich
Persian Gulf area is also underserved, especially Iraq. But in Africa there is “lots of talk, but few results”, according to Rawle.
Meanwhile, supply on transatlantic routes is tightening because no new submarine fibre has been laid on the route since 2002, when
Apollo (the system owned by Cable & Wireless and Alcatel) was commissioned.
All this investment will depend on global economic circumstances. Rawle is forecasting a fall in submarine fibre investment in
2010-12, though he believes it will pick up again after that. Though prices per bit will continue to fall, demand for Internet bandwidth is
strong. The Internet has not yet finished changing the world.
|
Press Room 
