Photons can't do it all-yet
Rick Barry, the chief technical officer at Sycamore Networks, recently told me that some of the most important breakthroughs in optical networking over the last 12 months or so have been electronic. Granted, Sycamore currently resides on the electrical switching side of the border that runs between transparent and opaque switching, but given the general drive toward all-optical networks and Sycamore's expertise in that area, it was novel to hear that electronics continues to play a significant role in optical systems design.
Then, two days later, I was in a conference room at Corning, listening to a presentation on the company's ongoing work in optical amplifiers. Intelligent amplifiers were the next big thing, said the speaker. Therefore, underneath all the pumps and circulators and other optical components normally thought of when conjuring an erbium-doped fiber amplifier, Corning is putting in a layer of electronics to perform the monitoring functions customers have come to demand. Of course, Nortel Networks has been touting its smart amplifiers for some time now. Again, the secret is electronics, not optics.
Thus, we have optical-networking engineers as interested in application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), and the new generation of network-processor chips as they are in liquid-crystal switching and arrayed waveguide gratings. The interest in this kind of technology perhaps seems logical for the optical-switch manufacturers, who are looking for the next step away from SONET/SDH-based networking (which, of course, has always relied on electronics for most of its functions). Until someone comes up with a practical way to read the entire signal content of a wavelength, the engines for packet, cell, and frame processing will continue to be electronic. In fact, one of the two advances Barry mentioned was SONET processors, which will enable boards to perform functions that previously required entire systems.
Even some of the most spectacular advances in optical networking have an electrical underpinning. Take ultra-long distance transport, one of the capabilities that made Qtera an attractive acquisition for Nortel. Soliton transmission and Raman amplification represent the optical technologies responsible for Qtera's success, but they wouldn't be as effective without an electronics-based boost from forward error correction (the second of Barry's big two-you didn't think I'd leave you hanging on that, did you?).
Of course, optical-networking companies will use these electrical components in their own unique ways. For example, Ted Rado, the director of marketing at startup Alidian Networks, tells me that the economics of optical equipment stand the usual electronic cost models on their ear. So while "traditional" electronic-based data networking gear might call for ASICs as a less expensive alternative to FPGAs from a manufacturing standpoint, the cost difference between the two technologies is lost in the noise when compared with the significantly higher cost of optical components. Thus, optical-networking companies might opt for the flexibility the software and firm ware -based FPGAs and network processors may provide over the ASICs that are dominant in other communications applications.
Thus, optical technology will continue to push copper-based electronic topologies to the very edge of the network (and perhaps into relative oblivion in the future), it won't mean that electronics itself will become less important. Those pesky electrons will continue to have a place in the all-optical networks of the future.
Stephen M. Hardy
Editorial Director and Associate Publisher
June 5 at SUPERCOMM
The Minority- and Women-Owned Business Enterprise (MWBE) will be holding free sessions at SUPERCOMM 2000, Monday, June 5. Session M05 (10 am to noon) is "MWBEs Excelling in a Convergence Environment," and Session M10B (2 to 5 pm) is "MWBE Executive Panel-A View from the Top" and "Meeting the SUPERCOMM Chal lenge in a Converged Environment." These sessions were listed incorrectly in the May 2000 issue of Lightwave, so please mark June 5 on your program schedule.