Every advancement in switch performance encourages carriers to push out even more bandwidth, and that puts more pressure on the installed switches, which eventually is feared to be unable to keep pace. The speed is all-important with every new technology trying to bring down another bottleneck. State-of-the-art telecom networks use light travelling through fibre-optic cables, instead of electrical signals travelling over wires, to carry voice and data. Optical fibre has seen dramatic increase in its traffic carrying capacity due to Dense Wave Division Multiplexing (DWDM) technology*.

This technology has been pushing up the number of channels that can be carved out of a single beam of white light, increasing rapidly from 20 to 40 to 80 and now to 128. There seems to be no upper limit. Each of those channels of light needs to be routed, switched, and multiplexed. The networking equipment slows down the traffic on a fibre network as virtually all the switching on fibre-optic networks till date is being done electronically. The "photons" of light that carry the information have to be turned into "electrons" or electrical signals so they can be read by the existing generation of network equipment. And after the switch has decided where a message is headed, the signal has to be converted back to light signal and put back on the fibre for the next stage of its journey. This conversion-and-reconversion process takes time, and that limits the performance. Therefore, the carrying capacity of such a network is determined not only by the size of the pipes it employs, but also by the speed of these switches. While light is dramatically faster than electricity, it is only as good as the switches that direct it.

The preferred approach would be switching without converting the signal from optical to electrical and back to optical again. This is important because, while electronics' capabilities continue to improve according to Moore’s Law (doubling every 18 months), optical systems have the potential for far more rapid evolution and improvement, and even greater capabilities*.

A new dawn in switches—and, in turn, network performance—is coming in the form of "optical" or "photonic" switching to forge ahead in its goal of driving more bandwidth onto backbones. Optical switches are expected to be in operation by the carriers before the ever-increasing rush of bandwidth overwhelms the existing generation of switches in the next three to five years.

The optical or photonic switch/router can intelligently switch/route wavelengths without any opto-electronic conversions. Optical switches would not reduce that bottleneck as these traffic routers would not need to turn photons into electrons and back into photons to manage the flow of messages. Instead, they would be able to find the routing directions for each message in the light itself, and would route light through optical cross-connects without electronics. That would speed up the flow, thus increasing the capacity of a fibre network. Besides the immediate cost savings that would result from that boost, it now looks possible that optical switching will simplify network architectures and cut down on the amount of network routing equipment that each fibre needs. The cost savings are projected around 50 percent for a network that has been reconfigured to take advantage of optical-switching architectures. The cost advantage for a newly built state-of-the-art system over an older system, that cannot be substantially reconfigured, is likely to be even larger.

Optical switching systems offer carriers dynamic management of their optical-layer network, rapid optical service provisioning and reduced operations costs. Carriers would be able to manage their multi-wavelength networks much more efficiently than is possible using the existing optical network gear. Thus, optical switching promises to be the "next big thing."

Carriers are embracing optical switching because it could give them and their customers a marketable advantage.

It is like being in the middle of a technological tornado. The fastest growing telecom business is the optical market, last year doing over $5 billion. Another reflection of it is the catapultic rise of valuation of the optical networking companies like Sycamore founded by Desh Deshpande, Cerent of Vinod Khosla (who later sold it to Cisco) …putting Indians in the forefront of the technology of the new millennium. And that includes Arun Netravali, head of Bell Labs, Pradeep Sindhu of Juniper, and Krish Prabhu, Alcatel, US.

One such router, Lucent’s WaveStar LambdaRouter, promises awesome capabilities: 256 ports—the capacity of 256 separate optical fibres, each carrying a single wavelength—each starting at 40 Gbps—pushing a total capacity of 10 Terabits. WDM allows each fibre to carry more than one wavelength.

The LambdaRouter uses "micro-mirrors" to switch lightwaves by reflection on a very miniature scale. The wavelengths tell the mirror how much to bend to route the light appropriately by using a "digital wrapper"—the optical equivalent of a packet header. In addition to switching the lightwaves, the wrapper contains information on restoration and the type of traffic being carried.

Another technology uses "inkjet printing technologies" to steer wavelengths. The basic chip is a matrix of 32x32 waveguides which are simply grooves in glass, filled with a liquid that allows light to pass easily. At each intersection of the waveguides, a bubble can be inserted. If the bubble is there, the wave cannot continue down the groove and is reflected off to one side or the other, switching the wavelengths.

The impact of optical switching could be as dramatic as DWDM*, which transformed core network transport almost overnight. Deshpande of Sycamore has predicted that technology was advancing fast enough and within 18 months, there will be enough bandwidth for the whole Internet business model to change.

A US study claims that spending on optical switching systems will run to more than $31 billion in Europe and North America over the next five years. The principal driver will be the rapid adoption of DWDM technology among new and established network operators.

Optical switching is the technology of the future, but really how far is the future. Undoubtedly, optical switching promises immense technological and cost advantages. However, it is not possible to throw out an existing network worth billions of dollars. Optical switches are already in the market, but to improve their performance and scalability involves a host of challenges, including deciding on the core technology involved. Therefore, the transition for established carriers may be slow and may be spread over five years. During that time, companies will try to add as much optical switching—with some electronics still being there—to their networks as they can, while preserving as much of their existing investment as possible. However, for the new players the attraction of being able to offer "customer controllable bandwidth"—with pricing based on use, network viewing and management capabilities—and becoming optical switched service provider will drive embracing of such networks rapidly.

Some say, the days of o-commerce (optical-commerce) are ahead!

by Niraj K Gupta for "Voice and Data" (From my Cell), August 2000.