The market for dense wavelength-division multiplexing systems has grown at a staggering pace. In just three years, the market has quadrupled from just under $1.7 billion in 1997 to about $10 billion in 2000. As the market for these systems evolves, more vendors are entering the business, wavelength-handling capacities are growing exponentially, and new products are being developed to address each segment of the network.
Even the contrast between the market in late 2000 and early 1999 is dramatic. In a May 1999 study, KMI Corporation (Newport, RI) identified approximately 15 vendors who reported sales of DWDM systems. As of September 2000, this list had grown to about 25--a number that would have been higher except for the quick acquisitions of some new players by other vendors--such as, Qtera by Nortel (Brampton, ONT, Canada), Chromatis by Lucent (Murray Hill, NJ), and Qeyton by Cisco (San Jose, CA). Fortunately for these competitors, the number of carriers deploying DWDM systems has grown faster than the number of vendors selling them. In its 1999 report, KMI identified approximately 75 carriers worldwide who had publicly announced deployments of DWDM. In just one year, the number of contracts announced by vendors has nearly doubled to about 140 in 2000.
Channels galore
In the early years of DWDM product development, channel count was a key differentiator among vendors. Manufacturers are still focusing on building up wavelength capacity, with channel-count capacity doubling about every year. For 2.5-Gbit/s transport, channel-count capacities have progressively risen from eight in 1997 to 16, 32, 40, 64, 80, 96, 128, and up to 160 this year. Lucent will offer a 320-channel product in 2001.
Manufacturers have increased channel counts in several ways. The first approach is to narrow the channel spacing on the filter. While filters with spacings of 200 GHz predominate in today�s market, 100-GHz filters are rapidly penetrating the market, and 50 GHz filters will be coming on strong--wavelength capacity doubles each time channel spacing is halved. Achieving such narrow channel spacing, however, can be expensive.
Another approach is to take wider-spaced filters and use an interleaver to achieve narrower channel spacing--effectively halving the channel spacing of the output wavelengths and doubling the channel counts. Most major vendors pursuing densely packed DWDM systems are using interleavers, providing a strong market in 2000 for component vendors such as Arroyo Optics (Santa Monica, CA); E-Tek Dynamics (San Jose, CA), which merged with JDS Uniphase (Nepean, ON, Canada) in June 2000; and WaveSplitter Technologies (Fremont, CA).
Yet another way to double channel counts is to open another realm of the spectrum. Amplifier manufactures have designed erbium-doped fiber amplifiers (EDFAs) to work in the long-wavelength band (or L-band). While conventional EDFAs work in the C-band (about 1530 to 1565 nm), L-band amplifiers open up wavelengths in the 1570 to 1610 nm range, thus doubling the number of channels carriers can transmit over a single fiber. Most major vendors have developed L-band products, which should begin to be deployed more widely in 2001.
DWDM, here and there
Amplifiers are also key in the development of new DWDM products for different levels of the carriers� networks. For the first four years of DWDM development, products focused almost exclusively on long-haul transmission that typically represented unregenerated spans of up to 600 km. The development of Raman amplifiers, along with forward-error-correction techniques, has stretched the distance over which DWDM can be transmitted without regeneration. First pioneered by Qtera (now a part of Nortel) and Corvis (Columbia, MD), these new ultralong-haul products have been or are being developed by most major systems vendors to reach distances of 3,000 km and farther. Combined with the development of intelligent optical cross-connects, these ultralong-haul products will bring about new network configurations for carriers� nationwide high-speed backbones, a mesh configuration with protection and restoration at the optical layer and the ability to provision and reconfigure traffic quickly from remote locations.
However, DWDM is also moving downstream in the network, ever closer to the end-user. For several years, DWDM systems vendors have been touting its benefits for metropolitan area networks, and we are just starting to see metro deployments this year. Nortel, Ciena (Linthicum, MD), and ONI (San Jose, CA) have had the strongest sales in this market to date. Initially, many metro deployments have been for point-to-point interoffice transport, an application similar to traditional long-haul DWDM deployment. True metropolitan optical networking systems, however, are being deployed in ring configurations with optical switches for restoration and protection at the optical level. While vendors have been promoting the benefits of metropolitan optical networks for revenue-generating wavelength services, many metro carriers--especially the former regional Bell operating companies--have been slow to implement DWDM metro rings, claiming the technology remains too expensive. Initially, metro DWDM will be attractive to data-intensive or fiber-poor competitive local exchange carriers or to metro carriers with a business plan based on wavelength services. As DWDM costs fall and the economic rationale proves itself, DWDM will become more pervasive on metro backbones.
Dense WDM also is moving into access and enterprise networks. For enterprises, the most common application today is storage area networks. Financial firms and corporations with large data needs are often leasing fiber to the data centers, and DWDM packs bandwidth onto their limited fiber resource. ADVA Optical Networks (Munich, Germany; Ramsey, NJ) and Sorrento Networks (Santa Monica, CA) have been the first and largest players in this area.
In products for access networks, vendors are focusing on delivery of a wide range of data and voice services to small or medium-sized businesses where cost--not channel count--is a chief factor. Most of these vendors are using passive optical networks to achieve multichannel transmission. So far, this market is small and may take a few years to develop, but it is being actively pursued by companies like LUXn (Sunnyvale, CA), Quantum Bridge Communications (North Andover, MA), and Terawave (Hayward, CA), among several other startups entering the market this year.
As DWDM moves closer to the end-user, the systems must accommodate a wider range of bit rates and protocols. A few years ago, the predominant interfaces for DWDM were OC-48 (2.5 Gbit/s) and OC-192 (10 Gbit/s) SONET transmitters. Metro and access networks now are introducing as low as OC-3 (155 Mbit/s) SONET interfaces into wavelengths, and enterprise applications may be using Gigabit Ethernet (GbE) or other lower-bit-rate data interfaces.
Because of the emergence of lower-bit-rate traffic on wavelengths, another trend emerging in the market is wavelength packing. Using an entire wavelength to transmit one OC-3 is highly inefficient. Several vendors, such as Centerpoint Broadband Technologies (San Jose, CA) and Sycamore (Chelmsford, MA), are developing products that aggregate these lower-bit-rate streams efficiently onto a single wavelength.
As the market for DWDM matures, DWDM technology is fading as a primary driver for product development. Like the chip to the computer, DWDM is becoming merely the high-speed mechanism for delivering services. Thus, DWDM transport products are focusing more on carrier service needs.
Based on the overwhelming demand for bandwidth and on carriers� needs to provide new revenue-enhancing services at lower costs, KMI forecasts continued strong growth in the DWDM systems market. Following a doubling of the market from 1999 to 2000, the consulting firm also expect DWDM sales to grow at a healthy but more moderate rate (35% CAGR) in the 2000 to 2005 time frame. The annual growth rate will decelerate to 20% by 2005, when the market will exceed $40 billion.
Neil Dunay, Senior Analyst, KMI Corp.