Service providers use wavelength management to propel optical networking
internettelephony.com
LIANE H. LABARBA
The facts are clear: Bandwidth demand is out of control, and service providers are clamoring to sate their customers' voracious appetites. The most obvious method of opening pipes is to turn up more fiber, often via wave division multiplexing.
But before service providers add lambdas willy-nilly, they must ensure that their management systems can capture the information necessary to support five 9s protection. For that, Sonet has been the traditional solution. Now vendors are challenging the need for optical-electrical-optical (OEO) conversion and devising methods of wavelength management that will push forward the still-distant goal of all-optical networking.
But similar to most things in life that we desperately long for, reaching that goal is never an easy process nor a fast one. The trick is for service providers to look past the current limitations, challenges and faults of today's wavelength management solution to the opportunities of the future. And while wavelength management technologies still are young, the need for standards and cooperation among equipment vendors may represent one of the most difficult yet attainable goals.
As networks become more of an amalgam of different vendor's equipment, interoperability between those vendors will be just as critical as the wavelength management technologies themselves.
Feed the need
The need for wavelength management is as clear as the demand for bandwidth, which resulted in the development of dense WDM (DWDM). But, optical networks are limited by management technologies.
When DWDM rolled across networks, it addressed fiber exhaust. But it falls short in the areas of lowering bandwidth costs, improving transport efficiency and increasing network reliability.
Previously, the only way to manage optical channels, provide signal regeneration and analyze and monitor signals was through Sonet. But a new era of wavelength management is emerging, and it potentially may offer carriers coveted new capabilities while keeping the functionality and reliability aspects of Sonet.
"[Wavelength management] will be very important for us going forward," says Ron Haigh, senior network architect for Qwest Communications. "The amount of demand for bandwidth on the customer side is skyrocketing, and this will give us options to satisfy those needs far differently than in the past. Customers will be more and more interested in [the] performance of the network than it the past, too."
Others echo Haigh's sentiment. "We are seeing a fundamental shift in how wavelengths are addressed," says Chris Nicoll, director of infrastructure analysis for Current Analysis. "It used to be that a wavelength was just a dumb pipe." That has changed with optical innovations from Lucent Technologies, Nortel Networks and Sycamore Networks, Nicoll says.
Giving life to the light
The problem carriers and vendors are trying to solve with wavelength management is no different from circuit or packet network management. In all those scenarios, service providers want to keep specific information intact as it travels across the network.
To provide the wavelength management, certain data is needed at all times. The information must be kept together, its location must be known at all times, its health should be determined and its destination must be known.
Several methods of wavelength management initially were created, but through experimentation and the process of elimination, only two inherently different methodologies surfaced. One method, called out-of-band monitoring, entailed the use of a separate, additional wavelength to carry the management function.
If a fiber carried 16 wavelengths, the proposal was to add a 17th wavelength to control the other 16, says Kathy Szelag, vice president of marketing for Lucent. The 17th wavelength would carry data such as information on where the wavelengths were going and their health.
"That method didn't win and was turned down primarily for two reasons," Szelag says. The first is that when out-of-band wavelength management was used, only the health of the 17th wavelength was judged; the health of the other 16 was merely inferred. The second reason is that if the optical power dropped 5%, for example, the separate wavelength could not determine if all 16 lost 5% of their power or if one of the 16 wavelengths actually was lost. The out-of-band method also presented problems because the 17th channel created abandoned signals.
Last June, the other top contender for wavelength management won out. Called in-band monitoring, it involves the use of a "digital wrapper." The International Telecommunications Union and the Optical Networking Forum supported the in-band monitoring technique, and a majority of carriers and vendors agreed that the digital wrapper should serve as the standard. As a result, vendors such as Lucent, Nortel and Sycamore are building to that standard.
In addition, the digital wrapper standard allows WDM systems to support more channel counts per fiber, higher bit-rate signals on each wavelength and greater distances between regenerators. In-band management gives providers more visibility into the network, which they can monitor on a per-wavelength basis (Figure 1).
By managing bandwidth at the wavelength level, service providers are not continually forced to overbuild for capacity, says Don Smith, vice president and general manager of OPTera solutions for Nortel. "The ability to understand the performance of the network, to isolate and restore faults, is critical. It gives service providers flexibility they have never had before."
In some cases, this translates to more wavelength-based services. Relative newcomer Quantum Bridge has developed a unique way to address wavelength management.
Based on the principles of passive optical networking (PON), Quantum Bridge's concept uses optical access networking to get rid of bandwidth bottlenecks. Optical access networking consists of an optical access switch at the central office, an intelligent optical terminal at the customer premises and PON between those two elements, says Jeff Gwynne, founder and vice president of marketing for Quantum Bridge.
"We are enabling dynamic wavelength slicing, which shares a wavelength among many customers," Gwynne says. With the dynamic wavelength slicing, Quantum essentially "fans out" each wavelength to deliver services to customers (Figure 2). That, in turn, provides a platform for a wider array of services and reduced costs, Gwynne says.
Piece by piece
While Nortel and Sycamore are integrating the wavelength management function into their existing products, Lucent is doing the same but calling it the WaveWrapper.
The standard, referenced earlier, calls for a digital wrapper to be added to each wavelength at its origination point. It travels with the wavelength through the network.
The digital wrapper contains a wavelength identifier and a destination identifier. Ideally, it also should contain information about the wavelength itself, such as its speed and protocol. By recognizing that a wavelength is 2.5 or 10 Gb/s or that it is an ATM or IP circuit, service providers can determine what tests to perform.
But despite the similarities of wavelength management to circuit or packet network management, it differs on the determination of health of the signal. Because the wavelength is a stream of light, it cannot be measured in the same manner as an analog or digital signal. If the light is present, the power of the network can be measured. If the light is not there, a fiber cut has most likely occurred and restoration is needed.
"We wound up having to add a very small digital header to each wavelength," Szelag says.
As the wavelength travels through the network, a variety of measurements are taken; the light passes through three to four different network elements that actually perform the management functions. The first element is WDM equipment, which could be a metropolitan or long-distance system. Transponders at the ends and amplifiers along the route check the health of the wavelength as it ventures through the network.
Routing wavelengths presents a gray area, however. If a wavelength is aware of its origination and destination sites, how does it reach its destination if it is not traveling on a point-to-point system? Two elements exist for this purpose: the optical cross-connect and the optical add/drop multiplexer. An optical cross-connect reads the wavelength "wrapper" and theoretically self-routes the light. Many optical cross-connects are under development, but most won't be available for about a year, Nicoll says.
Some optical cross-connects still have an electrical core. This raises questions about whether or not the management capability travels all the way through the device. "There are differing opinions, but in our view, the answer is yes," Szelag says. "Why would you take everything down to electrical in the optical cross-connect? We believe it should be all-optical, self-routed."
But vendors such as Tellium believe the signal must be converted to the electrical level in the cross-connect. Tellium's Aurora optical cross-connects actually regenerate the signals, boosting the power of the photons and demultiplexing them as they travel through the optical cross-connect. The Aurora cross-connects currently are used by Extant Communications and the U.S. government, says Nick DeVito, director of marketing for Tellium.
"The idea is to operate the network in the most efficient way possible," he says. "And we can provide that restoration."
Still, Szelag argues against OEO conversion. "There are valid reasons [for dropping back to electrical], but in the long run, as standards become more prevalent and the technology improves, you won't have to do either the [regeneration or the demuxing]," she says, noting that optical self-routing devices will be on the rise.
One of carriers' favorite all-optical promises is the reduction of regeneration sites, which significantly lowers network costs and the overall cost of bandwidth. Qwest, for instance, plans to enjoy a 70% to 90% cost reduction once it implements its all-optical network, which brings the number of needed regeneration sites from 30 to two.
"It's really a function of distance," Haigh says. "We try to eliminate as much of the electronics as feasible, but we still need it for grooming."
Chips in the glass
Electrical interfaces appear to have a firm position in optical networks for several more years. One reason is that wavelength management is in its infancy and can't compete with the survivability and reliability of Sonet.
"A lot more information is required in [pure] optical networks than what is required in traditional Sonet," says Mike Anderson, director of product marketing for network management at Sycamore. "Sonet's emphasis on protection and redundancy is still useful."
But Sonet seems to be evolving as wavelength management systems develop. "Sonet will continue to play a role, but it will be a different one," says Qwest's Haigh. "The notion of 'thin Sonet' will develop even further."
But addressing long-haul and metropolitan needs may be a tough spot. Wavelength management on metropolitan networks is fairly simple, and many vendors offer solutions in that area, says Lucent's Szelag. "But the problem arises when you go into big networks--not just the number of wavelengths, but geographic distances." The products capable of self-routing on long-haul links are not expected until the middle of this year, she says.
Not true, says Andy Wright, chief technologist for optical networking at Williams Communications. Williams uses Sycamore equipment to offer its "Optical Waves" service. "For us, the Sycamore equipment's distance capability exceeds the distance of our longest route, so [distance] is really not a problem," he says.
But the ability to perform management functions all the way across optical networks still is not fully addressed, says Nick Tanzi, president and chief operating officer of MFN. "We would really like to see an end-to-end view," he says.
Management visibility is crucial, Haigh says. "Section-by-section visibility is available, but it can't be seen from the edges," he says.
In addition, customers want some level of Sonet transparency so wavelength management doesn't effect specialized services. "People argue that management at the DWDM level is simplistic, but that has not been our experience," Tanzi says. "The tool sets to fault-isolate problems are light years behind the curve."
Complications with wavelength management also arise with the frequent switching of traffic between long-haul and metropolitan routes, he continues. "There are a lot of challenges in hops off long haul and hopping onto metro," Tanzi says.
Managing wavelengths across multiple networks is only one of many challenges. "[Developing] operational systems to support multivendor networks seamlessly, so [wavelength management] doesn't matter from a network management perspective, is another concern," Haigh says.
Loss, particularly as wavelengths travel through optical cross-connects, is another problem optical cross-connect vendors must tackle, Szelag admits. "The [optical cross-connect] technology as a whole still has some issues with being too high in loss," she says. "If you were in a geographically huge network and you were going through multiple [optical cross-connects], you would have more loss than you would want."
But while wavelength management technology is more advanced in point-to-point and ring-based architectures, the technology is less developed in mesh architectures. The larger the mesh design, the more complex the routing becomes. That, in turn, causes delayed restoration. Signals sometimes are regenerated unnecessarily to boost the power.
The jury still is out on building generic [pure optical] mesh networks with the restoration times of Sonet, Haigh says. "Vendors are making great strides [in that area] from a technology perspective though."
Where to go from here?
Although wavelength management still must address numerous challenges, it clearly has captivated service providers, which recognize its potential. Their piqued interests fueling experimentation, trials and rollouts.
The questions lie in how the issues will be resolved and how long it will take to solve those problems.
Many believe the recent formation of the Optical Domain Service Initiative, spearheaded by Sycamore and boasting more than 50 vendor and carrier members, may be a critical step in attaining end-to-end optical network management.
"ODSI has the most promise for ushering in interoperability and all-optical networks," Nicoll says. "Carriers want everything to be best of breed, and ODSI will hopefully help them get that."
Within the next year, Tanzi expects to see wavelength management reach its next level and address operation support system integration, the ability to switch between long-haul and metro networks, and other issues. "We will really start seeing the fruits of our labor then," Tanzi says.
--------------------------------------------------------------------------------
Wavelength management for sale
Liane H. LaBarba
Although wavelength management is a significant step to the all-optical future, most agree that electronics will remain in the optical network for many years to come.
Although wavelength management is still in the innovation Petri dish, Williams Communications is jumping in with its Optical Wave service. Using Sycamore Networks' equipment, Williams is providing unprotected optical wave services to help fill the gap between dark fiber and application-oriented capacity services.
Typically, dark fiber customers purchase fiber, deploy optronics and manage their own networks; dim fiber customers purchase dark fiber, optical amplifiers and regenerators and maintain their systems. With Williams' Optical Wave service, customers can have fully managed ATM and IP services without having to deploy optical equipment or manage their own networks, says Andy Wright, chief technologist for optical networking at Williams (see figure).
"We found the right balance between enough and too much service," says Eddie Garrott, life cycle product manager for Optical Waves at Williams. "There was too much flexibility in offering variable bit rates, so we found the middle ground and are offering OC-48 service."
In addition, the transponder-based system enables optical waves to use a single channel rather than the work and protect channels demanded by traditional Sonet. For performance monitoring, customers use a Web-based interface to view statistics such as bit error rates, equipment alarms, network availability and optical power levels.
Williams plans to augment the service, too. "We want to ramp up with the integrated network management piece into the OSS," Garrott says, adding that OSS integration with the Sycamore equipment has gone smoothly.
The key is that management of the system can integrate across a variety of architectures, says Mike Anderson, director of product marketing for network management at Sycamore. "Before service providers had to be aware of topology. That is changing."
Back
--------------------------------------------------------------------------------
For questions or comments,
contact Web Editor Karen Murphy msblues@earthlink.net .
Please report any problems with this site to:
Webmaster@internettelephony.com.
www.internettelephony.com
¸2000 Intertec Publishing Corp., a PRIMEDIA company
All Rights Reserved. |
