DWDM has what it takes to fill up the MAN holes

nyone who's taken a cross-country road trip knows how it goes: You speed along on the interstate highways until you reach a major metropolitan area. Unless you want to bypass the city, you find yourself dumped onto surface streets, crawling along in traffic until you reach your hotel or destination.

The high-capacity optical-fiber infrastructure enabled by dense wave-division multiplexing (DWDM) closely replicates that experience. DWDM, first implemented by Lucent and made commercially successful by Ciena, is a way of expanding the capacity of long-haul fiber by using multiple colors of light on the same fiber. Each color is its own independent channel and can be used for any type of digital traffic - voice, data, video, ATM or IP. Voice and data move cross-country at the speed of light through these optical-fiber channels until they reach the metropolitan-area network (MAN). Like the interstate, DWDM stops at the city limits.

In this decade, the DWDM industry has pushed the capacity of a single fiber from a couple of channels to almost 100. This has been great for long-haul carriers that need to address the boom in data networking.

It turns out, however, that the major carriers have been much better at laying fiber than they have been at finding applications that would use the immense capacity created by DWDM. I have seen estimates that only 5% of this fiber is actually "lit." The rest is "dark fiber" - fiber buried in the ground and held in reserve for the day when demand requires it. To generate demand, ways need to be found to bring this incredible bandwidth into the short-haul metropolitan ring, where it can be used by businesses and residences within the local loop.

Local carriers have deployed very little SONET and related enhancements such as DWDM because of the complexity and cost of the technologies. The same million-dollar Ciena DWDM transport system that is cost-effective for sending long-haul traffic from San Francisco to Chicago doesn't cut it economically when the job entails sending traffic from one building to another in the same city. That could change once MANs - the much talked about last mile to a local business or residence - are fully connected.

What we need are cost-effective, easy-to-install network products for adding, dropping and cross-connecting DWDM networks to MANs.

A number of start-ups are working to help move DWDM into the local loop. Cambrian Systems, in Kanata, Ontario, has tested linking its DWDM MAN equipment with Gigabit Ethernet pioneer Packet Engines' routing switches. Tellium, an Edison, N.J.-based spinoff of Bellcore funded by Ortel and Science Applications International, has won a contract to install a WDM optical-ring network for linking several government agencies in Washington, D.C. (Mayfield Fund has no investment in these companies.) The next generation of DWDM equipment could go beyond simply adding more point-to-point capacity. By creating subnetworks within the MAN, provisioned and managed on a wavelength basis, virtual private networks could be created on demand.

The recent failure of the $7.1 billion merger between DWDM vendor Ciena and Tellabs is more of a commentary on the vagaries of the stock market and Lucent's clout than of the future of DWDM. Companies such as Qwest Communications and Level 3 Communications have done a great job of putting DWDM in place. The real benefits from DWDM should occur within the next year, when the MAN holes begin to be filled and DWDM is finally brought to the locations where it can be used by businesses and consumers.

Fong is a general partner of Mayfield Fund, a venture capital firm in Menlo Park, Calif. He can be reached at kfong@mayfield.com.