Dave Rusin is out to shine a light on what he considers "one of the little dark secrets of the telecom industry."
According to Rusin, CEO of American Fiber Systems Inc. (www.americanfibersystems.com), the all-optical switches that have been such a hot topic in recent months won't work on the majority of fiber in the ground today.
"The idea of all-optical [networking] is to switch wavelengths in real time," notes Rusin of AFS, which provides dark fiber in second- and third-tier markets. "That requires a photonic switch. But fiber needs to allow that. So you need fiber that is full-spectrum fiber. Lucent AllWave fiber is what we use."
Extolling the benefits of the AFS's state-of-the art fiber network is part of Rusin's sales pitch. He's been telling the CLECs they need to get off incumbent carriers' infrastructure because "it's old and because it's money in [ILECs'] pockets." At the same time, the company is hawking its glass to ILECs, suggesting they buy rather than build.
While there is some merit to those arguments, other optical specialists in the industry refute Rusin's claim that all-optical switches can't work on existing fiber plant. Techniques such as dispersion compensation and specialized lasers can allow single-mode fiber to support lambda switching, but questions remain as to the economic viability of those fixes.
"If you look at long-haul network fiber, like Lucent's TrueWave and Corning LEAF or generically SMF-28 fiber [single-mode fiber], those fibers are optimized for fixed wavelengths," Rusin says. "They can operate at 1300 to 1550 nanometers." Spacing between wavelengths at 1300 to 1550nm, he adds, is typically 125 to 200 GHz windows today.
"But to switch wavelengths, you need a broader spectrum," according to Rusin, "so the characteristics of the glass have to allow that." That means network operators that use all-optical, aka lambda, switching need to have fiber that allows traffic to travel over a broader spectrum. They also need fiber with smaller windows (about 25ghz) between wavelengths to allow for real-time switching between wavelengths to allow for on-demand bandwidth provisioning for services like video, Rusin says.
"You may need a lot of bandwidth for video at one minute and less the next minute because [at that point there's] maybe just voice," he continues. "That's a different nanometer window, and to accommodate that, glass needs a lower water peak, with fewer hydrogen ions." A lower water peak means less dispersion in the glass.
Janice Haber, systems development and strategy vice president with the network products group at Lucent Technologies Inc. (www.lucent.com), says her company has led the way in introducing "application fiber," referring to fiber designed for specific applications, such as long-haul connections. The company's TrueWave RS, a non-zero dispersion fiber for which every channel has the same pulse, is best for long metro distances where single-mode fiber can be a problem because of the amplifiers that would be required every few kilometers to address loss. (Amps are expensive to purchase and can be expensive to maintain.) For metro, local and access traffic, AllWave is the most economical and upgradeable of Lucent's fiber products, she says, adding that AllWave is selling up to five times faster than TrueWave, which itself had "hockeystick growth." And at SUPERCOMM the company unveiled a product integrating TrueWave and AllWave.
At 10- to 40-gig single-mode fiber limits carrier reach to about 60km "and that's if it's running really well," Haber says.
Another problem with today's widely deployed single-mode fiber, Haber adds, is polarization mode dispersion. She explains that single-mode fiber has two electronic modes and that if a fiber is not circular or is bent, polarization mode dispersion occurs, which means traffic gets out of sync because the electronic modes are not working in concert. "When you try to correct for it, it's very expensive," Haber says. "You need modes together to switch the whole lambda."
Haber adds that incumbent IXCs haven't put out much new fiber. When those companies want to move from 2.5gbps WDM to 10gbps, she adds, "they'll lose the economics, so they will have to think about new fiber." Some carriers, like Level 3 Communications Inc. (www.level3.com), had the forethought to install large enough ducts that they will be able to simply feed fiber through the underground pipes when needed, says Haber. But others might not have the underground assets to accommodate space for new glass, so they might have to dig or consider leasing fiber to meet their needs.
While no one will argue that using the latest "application" fiber is the best way to go, several vendors in the optical space told xchange that there are various ways to compensate for the shortcomings of older fiber that's in the ground. But the affordability of these methods remains to be seen.
"Fiber tends to be permanent, and almost anything you can do with lasers is cheaper than purchasing new fiber," says Floyd Ferguson, director of strategic marketing at Fujitsu Network Communications Inc. (www.fnc.fujitsu.com). He adds that Fujitsu can engineer products like its FLASHWAVE OADX DWDM product for individual fiber types, which may require the use of amps. Ferguson explains that systems integration is a huge part of the optical equipment sale, adding "This isn't plug and play like SONET."
Curt Weinstein, director of metro fiber for Corning Inc. (www.corning.com), agrees that using a different kind of laser can help prolong the life and usefulness of embedded fiber. Externally modulated lasers, which are complicated lasers with very high performance, might be used to support lambda switching in the future, he says. These lasers are already used ubiquitously in the long haul, he says. But it remains to be seen if they will be used in metro applications, he adds, saying it could be too expensive.
A second option is to employ dispersion compensation, which means adding amplifiers on the fiber. It works like this: The dispersion compensation device has negative dispersion, while single-mode fiber has positive dispersion. So they cancel each other out.
But dispersion compensation also has its problems. According to Weinstein, the cost of adding amps on single-mode fiber in the long haul is prohibitively expensive. And the fact that metro applications can be in single or multiple rings, or meshed configurations makes fiber amplification much more challenging than in point-to-point applications.
"Cost varies by size and topology of the network," says Weinstein of Corning, which, like Lucent, offers specialized long haul (LEAF) and MetroCor fiber, which he says will support lambda switching. "But clearly the vast majority of fiber is standard single mode," he adds.
But Richard Cunningham, senior analyst with Cahners In-Stat (www.instat.com) sees the limitation of single-mode fiber only as "an irritant" to the adoption of all-optical networking.
"Granted, a lot of fiber in the ground that predates 1996 or so isn't fully usable, but that's a declining issue" as carriers upgrade their networks with newer fiber, he says. "A bigger issue is getting enough fiber in the loop, which many companies are working on, but which cities are increasingly sensitive about," he says, noting many cities' resistance to carrier requests for digs to install new fiber and fiber conduits.
