The future of optical networking
SPECIAL REPORTS / Optical Networking/WDM
Providers need to boost higher bandwidth for Internet users while lowering their operational costs.
KEVIN AFFOLTER, Agility Communications
Delivering bandwidth at a profit is one of the most pressing challenges service providers face. Providing optical-networking service requires the business to forecast demand and commit to major financial expenditures based on these estimates. Today's extraordinary bandwidth requirements require the communications industry to look for new solutions that maximize the capacity of all channels on a network, reduce the network's complexity, and lower the costs of delivering service. Fortunately, the breakthrough technology of DWDM and the emergence of tunable lasers hold out the potential for pioneering service providers to increase fiber capacity and efficiency while simultaneously cutting costs.
Traditionally, fiber-optic networks are largely static in nature and provisioned using fixed-wavelength lasers with no capability to flexibly or dynamically provision traffic without going through an electronic fabric. It has become increasing difficult for the optical-networking system vendors to forecast how much capacity carriers-and their customers-will require, and therefore, the commitments made to purchasing lasers with a fixed wavelength can create delays in turning up services.
These delays can be especially serious when last-minute changes in network deployments change the wavelength requirements, leaving the business stranded with an expensive fixed-wavelength laser that it can't use. To counteract this forecast uncertainty for both volume and mix of fixed-wavelength lasers, inventory has been piled high at all stages of the supply chain-one of the reasons why the recent downturn has hit the industry so hard.
Today, thanks to DWDM technology, each optical fiber that used to carry just one optical signal (channel) can now carry in excess of 100 channels, delivering a huge increase in network traffic capacity of both long-haul and metropolitan networks without the massive infrastructure investment of laying new fiber-optic cable-not to mention protracted negotiation for rights-of-way. At first glance, the innovation of DWDM would seem to make a simple and perfect story of supply and demand or problem and resolution, but the reality is neither so simple nor so perfect. Most implementations of DWDM cannot consistently operate at their full theoretical capacity. Businesses move, people change Internet service providers, traffic patterns shift, and problems arise because although a DWDM network can carry many channels, it can't simultaneously carry two signals on the same wavelength.
The optical-networking industry requires a more flexible laser technology that can efficiently use the attributes of DWDM networks. Today, there are tunable lasers being introduced to the market that can tune across 20 to 30 channels. Although these lasers largely solve the supply problem, they do not enable system vendors and carriers to benefit from the operational and capital savings associated with the enhanced, dynamic optical-networking architectures that are now being introduced.
Though demand has taken a downturn in 2001, most industry observers expect it will pick up again-in a hurry-especially in the metropolitan networks. To support this future demand, laser technology is needed that can support these new architectures as well as longer distances.
As Internet users log on by the millions, there's still hope. Most of the limitations of DWDM can be addressed through the widespread deployment of widely tunable lasers, an emerging technology with the potential to enable bandwidth-on-demand for the benefit of Internet users around the globe. For example, when a business purchases a tunable laser, it's not committing to just one wavelength, so it no longer needs to forecast demand for more than 100 wavelengths. Instead, the business can purchase a single, widely tunable laser to cover the whole C-band and another to cover the L-band.
Likewise, traffic fluctuations no longer result in lasers having to stand idle and fiber-optic cable having to carry less than its capacity. Tunable lasers enable businesses to use this untapped bandwidth, with software rapidly changing laser channels from one wavelength to another to avoid wavelength roadblocks. Dynamic wavelength provisioning enables vast improvements in bandwidth use, leveraging the full capacity of the network to meet peaks in demand.
Widely tunable lasers that can tune across 100 channels are now being introduced to the marketplace, and although industry experts have endorsed high-power, widely tunable lasers as the way to maximize band width usage, only a handful of companies in the industry have the qualifications, facilities, and in-house expertise to manufacture these sophisticated lasers. Unfortunately, many of these companies have had extreme difficulty in scaling their development and manufacturing operations to meet the seemingly insatiable demands of the large optical-networking infrastructure players such as Lucent Technologies (Murray Hill, NJ) and Nortel Networks (Brampton, ON).
In addition, widely tunable laser technology will initially cost more than fixed-wavelength and narrowly tunable lasers, but these direct cost comparisons are misleading. First, widely tunable lasers are less mature and are not being shipped in the same volumes as their fixed-wavelength cousins. As production volumes increase, the economies of scale will prevail and the gap will be closed.
Secondly, with the appropriate flexible network architecture, widely tunable lasers can carry traffic virtually all the time, while the networks built with fixed-wavelength or narrowly tunable lasers, have less or no ability to reconfigure in-line with the dynamic changes in traffic patterns. DWDM systems that use widely tunable lasers recoup their additional investment many times over-effectively reducing the cost of bandwidth transport.
While service providers will not often disclose details on the amount of time they are unable to use bandwidth because of channel conflicts, industry watchers estimate that fixed-wavelength lasers are being sidelined for a significant and growing portion of their potential operating time-in some cases, greater than 70%. More and more frequently, last-minute changes in network deployments are leaving system and component vendors stranded with fixed-wavelength lasers they cannot ship. Additionally, the service providers are unable to light the service. Ultimately, this increases the per-unit cost of bandwidth that the service provider is able to sell.
High-power, widely tunable lasers can eliminate the problem of having some fibers under-used, while others are overloaded. For optical-networking carriers, the greater cost of purchasing tunable lasers is increasingly offset by the fact they will no longer have to forgo bandwidth revenue due to the shortcomings of fixed-wavelength lasers. With widely tunable lasers, companies can address a shift in traffic patterns with an automated, fast, and inexpensive command versus a truck roll to deploy a new laser.
Widely tunable lasers can also dramatically reduce the total cost of supplying bandwidth, virtually eliminating the expense of "sparing." In networks deploying fixed-wavelength lasers, each laser needs to have a dedicated spare. The spare is held at the circuit pack level-each costing tens of thousands of dollars-resulting in millions of dollars tied up in spare inventory.
Many carriers have managed to decrease this investment significantly by sharing spares between sites or assigning a dedicated spare wavelength so traffic can be switched to that wavelength when there is a transmitter circuit pack failure. In this case, the wavelength-specific spare switch-out can be scheduled into a maintenance window to minimize network impact. There's no avoiding the fact that the cost and space required for the sparing mountain is massive and, until now, is a necessary but wasteful use of company resources.
Tunable lasers offer a highly cost-effective alternative to traditional sparing practices. To fully realize the economies of this new approach to sparing, tunable lasers ought to be used widely throughout the network, as this results in fewer circuit packs that must be supported on the same optical-networking platform; it also drives the economies of scale for the laser vendors. But even using just a handful of widely tunable lasers for sparing still enables companies to eliminate an inventory of several thousand spare lasers. The cost savings are tremendous as businesses can then free up these dollars for more productive investments.
Carriers can further increase network revenues if they can offer the capability to dynamically add and drop wavelengths at more remote sites throughout a network (see Figure 1) so they can respond to changes in traffic patterns at network pinch points. However, with fixed-wavelength lasers, reconfigurable optical add/drop multiplexer (OADM) sites are not practical or cost-effective. These sites require the use of back-to-back multiplexers, switches, and a laser for every wavelength that may be added and dropped. That is clearly very costly and far from elegant. Widely tunable lasers, used in conjunction with a tunable or reconfigurable OADM, enable providers to take advantage of these new revenue opportunities.
Other applications for widely tunable lasers include wavelength conversion at photonic crossconnects where optical rings meet. That enables traffic to hop from ring to ring, without going through an ADM, by eliminating wavelength blocking (see Figure 2).
Fixed-wavelength lasers are still far more common than tunable lasers, but as service providers become more aware how much these lasers cost in forgone opportunities to supply more bandwidth, they are increasingly deploying tunable lasers in networks throughout the world. One of the most promising tunable laser technologies leverages a dual sampled-grating distributed Bragg reflector design to achieve very wide tuning from a single waveguide output (see Figure 3).
Linewidth and modal purity of these devices can rival the best commercial distributed-feedback lasers. Manufacturers can also choose to monolithically integrate high bit-rate modulators and amplifier functions on the same platform. With a wavelength range of up to 40 nm, these lasers can cover either the full C-band or full L-band with support for channels on 50-GHz spacing and high output power.
Industry watchers believe tunable lasers are well positioned to dominate the market. However, manufacturers must first break through the difficulties of turning out reliable products in high volume. Optical-networking companies will demand tunable laser vendors to demonstrate the same reliability they have come to expect from fixed-wavelength lasers. Also, widely tunable lasers will require availability from multiple sources, enabling businesses to improve security of supply before the technology is widely adopted.
Optical networking will benefit even more from tunable lasers when systems companies evolve infrastructures to leverage the dynamic capabilities of this new technology. Today, in the absence of such architectures, most service providers use tunable lasers to ease the planning problems associated with forecasting thousands of product codes and for reducing the number of lasers they have to hold as spares. As a result, only a fraction of the cost savings associated with deploying tunable lasers throughout their networks is realized.
However, companies with foresight in both the long-haul and the metropolitan markets are building new systems with racks of equipment and software controllers designed for lasers that can tune to any wavelength. As these systems come online, and as organizations replace fixed-wavelength lasers with widely tunable lasers that can tune dynamically, fiber-optic networks will be able to fully realize the potential of dramatically increasing bandwidth supply for metropolitan and long-haul traffic.
Kevin Affolter is director of marketing for Agility Communications (Santa Barbara, CA). He can be reached via the company's Website, www.agility.com.