Rough times call for rough technologies. As such, coarse wavelength division multiplexing seemed a perfect fit for the bumpy ride of the current telecom slump. As a cheaper, shorter-range version of dense WDM, it gave carriers a less expensive way to turn up new sites and new customers. And because it contains cheaper, less precise componentry, it provided a low-end product for metro DWDM vendors to turn to as sales for their high-end gear thinned out.

“They're really fairly dumb devices,” said Cogent Communications CEO Dave Schaeffer. Cogent has been using CWDM for more than a year now, and CWDM gear from Cisco Systems has come to comprise more than 80% of its deployments. CWDM can typically undercut the price of deploying DWDM by about 30% to 40%. But because Cogent's network architecture had been using DWDM gear somewhat inefficiently in its edge networks, the switch to CWDM cut Shaeffer's costs by 80%.

Saving that kind of scratch sounded anything but dumb to carriers, and vendors quickly fell in line, making CWDM one of the most talked-about optical technologies of the last 18 months. Though the total number of vendors currently offering CWDM is unknown, it's easy enough to list at least 15, from big names such as Nortel Networks, Ciena and Alcatel to smaller players like Lumentis and LuxN. Tellabs and Movaz Networks both threw their hats into the ring at the Optical Fiber Communication show in March, and Appian and Adva are showcasing new CWDM wares at Supercomm.

“Seems like every other day there's another CWDM player in the market,” said Mark Lutkowitz, Telecom Pragmatics analyst.

Unfortunately the market may not be big enough to support them all. The worldwide market for CWDM in 2002 was just $19 million, according to IDC — a mere 4% of the $483.8 million metro WDM market. CIR projects revenues from CWDM equipment to reach $63 million this year, growing to $240 million by 2007, but other analysts aren't impressed. RHK hasn't deemed the space worthy of report so far. And a spokeswoman for the Dell'Oro Group said the firm is “keeping an eye on” the CWDM space but not “tracking” or researching it “because it's so small.”

Given those figures, CWDM might be slim pickings even if the market share were evenly distributed among all its new players. But analysts say the lion's share will likely go to leaders like Adva, Nortel and Ciena, leaving the others underfed and famished.

“For all these metro core DWDM start-ups to all of a sudden start offering [CWDM] is sort of just following the herd,” said CIR analyst David Gross. “Some vendors have gotten a little too excited by what they've read in [requests for proposal]. Most of what gets written up in RFPs never makes it to the network.”

Whereas DWDM uses pricey, precise filters to pack up to 40 wavelengths closely together on a single fiber without interfering with one another (even more in the case of long-haul applications), CWDM uses fewer wavelengths (typically four or eight but in some cases 16) and spaces them far apart to avoid interference. This way the filters don't have to be so exact (or expensive). CWDM is also limited in general to distances of 40 km to 60 km, and its bands lie outside the range of common amplifiers, ruling out amplification of the signal — another limitation DWDM doesn't have.

So far there's been a lot of interest in using CWDM both in point-to-point enterprise applications (for businesses with large bandwidth requirements for data storage extension, for example) and as a way to connect new sites to the metro core, where its traffic can be groomed onto a DWDM ring.

Many vendors are marketing CWDM as a stepping stone to DWDM, a low-risk way for carriers to dip a toe into tall shiny buildings or enterprises. When customer bandwidth needs escalate, the story goes, carriers can then upgrade their C to DWDM, turning up capacity once the cost has been justified. But some say that scenario is likely to be rare at best, as only a fraction of businesses ever need more than the eight wavelengths CWDM provides.

In the meantime, competition in the CWDM arena could be cutthroat. Because most vendors are buying somewhat commoditized, off-the-shelf components from the same clan of suppliers, there's little room for competitive differentiation. And because CWDM's chief selling point is its cost-shaving quality, carriers are most interested in the upfront “first-in” expense anyway. As a result, vendors end up competing on price alone, an unfavorable position for any sales organization.

“Cost competitiveness will be key,” said Sophia Fang, vice president of business development for Movaz, proving her own point by referencing price as a competitive advantage: “I can't see Nortel bringing the cost down enough to compete. Their basic platform to start out with is more expensive than most people's.”

Nortel, for its part, counts its experience with carrier networks and relationships with leading data storage providers such as EMC, Sun Microsystems, IBM and HP as competitive advantages, both of which describe Nortel as a company rather than its CWDM gear.

Nortel, which has announced SBC Communications as a CWDM customer, sells CWDM and DWDM in separate boxes on the same platform (its Optera metro 5100 and 5200, respectively). In contrast some vendors such as Tellabs and Lumentis are marketing systems that can use CWDM and DWDM together.

In general, using the two in concert is “interesting” but not necessarily compelling by itself, said Mike Guess, chief operating officer of wholesale carrier OnFiber, whose use of Ciena's non-concert CWDM gear is increasing after having deployed it for more than a year. When using the two together, “you have to design the system to the lowest common denominator because they're sharing a fiber pair,” he said. “I can't put amplification on the system and create optical add/drop rings. So I can't really build networks out of CWDM, but I can with DWDM.”

There are other means vendors could potentially use to distinguish themselves rather than just repackaging components into ever-smaller boxes. For instance, as a general rule, diagnostic tools for CWDM gear pale in comparison to that for DWDM systems, said Cogent's Schaeffer. As a result, testing and turn-up require a little extra effort, and troubleshooting takes a little more time, all of which adds cost to the equation. A vendor with better power monitoring and diagnostics could potentially offer a simpler solution and still claim cost benefits.

A vendor could also distinguish itself by selling bidirectional CWDM systems that use one fiber instead of two, Schaeffer suggested. “There could be places where you only have one spare fiber to use. Today you can't use CWDM on that particular ring. There's no technical reason why.”

Schaeffer also urged vendors to make the leap from the 2.5 Gb/s capacity of most contemporary CWDM gear to 10 Gb/s, a notion seconded by OnFiber's Guess.

“I'd love to see 10 Gb/s, both [Ethernet] and Sonet. Nobody's doing it,” said Guess. “We have customers buying 10 Gb/s waves. I don't have the choice to put it on CWDM, even if that were the appropriate solution based on fiber and distance. I have to put a DWDM solution in there, with all its overhead, even if I'm just talking two channels.”

However, at the moment, both 10 Gb/s and bidirectional systems would probably be small niches inside an already small market. Schaeffer admitted the need for bidirectional systems would only come up in Cogent's case about 10% of the time, and its added expense would cut into the cost benefits of CWDM.

Enterprise router vendors are expected to eventually add 10 Gb/s CWDM plug-in modules to boost the speed of their equipment. (At press time, the usual suspects in that group — Foundry, Force10 and Extreme Networks — did not return calls seeking comment, and a Cisco spokesman said the company would not comment on future products.)

When they do, it will further erode the validity of the CWDM-as-a-stepping-stone story because with all that capacity, customers will have less need to upgrade to DWDM. It will also shift much of the CWDM equipment market away from service providers and toward enterprises, which will account for the bulk of the growth CIR projects in the market. And it will also make the crowded CWDM equipment space even more crowded.

“There is going to be a heavy focus on CWDM. In the foreseeable future, I wouldn't be surprised if there were much more CWDM deployed than metro DWDM,” said Lutkowitz. “But there probably are way too many [vendors].”