By Robert Pease
The optical-crossconnect (OXC) market has undergone a significant transformation over its brief history, particularly as manufacturers add functionality and ports. Originally, OXCs were meant to simply connect paths of light and have relatively few ports. These devices were designed to enable reconfiguration of the network to accommodate different sets of circumstances and, at least until recently, were not considered intelligent or network-aware elements.
Conversely, optical switching has been considered a more dynamic way of establishing a network. However, this class of equipment has been unaware of signal content and has operated analogously to circuit-switched network gear. The recent addition of intelligent attributes to both OXCs and switches is making it more difficult to distinguish the two.
When you toss in the more intelligent routers that actually perform inspection functions on each packet or cell flowing over the network, the lines traditionally separating these individual markets tend to blur even further. Manufacturers are continuing to mix and match necessary functions into a single box, labeling it any of the above and adding to the confusion. A case in point is Lucent Technologies' introduction of its LambdaRouter, considered by many analysts and competitors not to be a router at all, but an all-optical crossconnect.
"Some of this is just semantics," says Lawrence Gasman, president of Communications Industry Researchers Inc., a Virginia-based analyst and consulting firm. "All-optical switches today are really crossconnects. However, since both switches and crossconnects actually 'switch,' a crossconnect can be thought of as a special kind of switch. A router is really a software device for moving packets through a network. A switch is a hardware device. Again, ignoring the optical thing for a moment, this definition is becoming a bit vague because routing and switching are being combined increasingly into one box."
Gasman also believes that within the optical space, optical devices are hardware by definition. Therefore, they are really not routers in the classical sense. That said, big router vendors like Avici Systems Inc. (Chelmsford, MA), Pluris (Cupertino, CA), and IronBridge Networks (Lexington, MA) are developing regular routers with direct optical interfaces. In fact, IronBridge says it will eventually build an optical-switching core into its product.
It's all very confusing, but the blame can be placed solidly on the fast-forward pace of optical technology. Combining the functionality of traditional crossconnects, switches, and routers into a single device that costs less than individual boxes is making a lot of people wealthy these days. Whether you label it a crossconnect, switch, router, or any combination (e.g., switch-router), customers just want to do more with less equipment at lower costs. Therefore, some of the trends in the crossconnect market are pressing toward that goal.
OXCs are being manufactured with more ports and more network-management capabilities built-in. Point-and-click provisioning is also a key feature requested by customers. Carriers are embracing truly intelligent (not just automated) systems that can rapidly provision wavelength services in a carrier-class system design. Crossconnects with the ability to offer SONET-level reconnect times are just around the corner.
"The current role of static optical crossconnects is to provide the physical connectivity for a network that does not normally have the restoration requirements addressed by SONET," says Bill Kleinebecker, senior consultant for Technology Futures Inc. (Austin, TX). "An example of such a network would be an all-Internet-protocol [IP] network where loss of path is handled by the IP router. Since a non-SONET physical network can operate at greater bandwidth at a lower cost, there is good value to be achieved from the optical-crossconnect market."
The current trends point to the ability to handle higher bandwidths and achieve less signal loss at interconnect points, adds Kleinebecker. As Multi protocol Label Switching (MPLS) and differentiated service protocols are implemented, the result is greater use of OXCs in lieu of SONET because traffic can be better shaped to take advantage of the optical-interconnect part of a network.
"However, by implementing MPLS and differentiated services, the stage for the next generation-optical switching-will be set," says Kleine becker. "So this stage will be relatively short-lived. We're already seeing this next generation in the form of opaque switches from companies like Cisco [Systems Inc.-San Jose] and Nortel [Networks-Brampton, ON]."
At Alcatel (Paris), the OXC market is viewed as a focus on "pure" optical solutions. The system, says Kent Novak, Alcatel-USA's vice president of marketing for transmission systems, should have a photonic matrix versus an electrical matrix, evolving toward the "optical 3R"-reshape, retime, and regenerate-when possible.
"Customers are seeking added intelligence in the optical core, which will require communication between the optical crossconnect and other optical equipment such as optical routers," says Novak. "Predominantly, optical-switch components are becoming more prevalent and more sophisticated, allowing for larger, more cost-effective solutions. At a higher level, protocols are being established that will allow the optical crossconnect to communicate with high-speed routers, thus enabling dynamic bandwidth allocation in the optical core."
Carriers are facing a dilemma of their own. With huge investments in legacy equipment, they must decide which will be ultimately more economical: Buy new products that will interoperate with aging equipment or yank out the old for the latest and greatest that offers scalability into the next decade and beyond.
"Two years ago, nearly all carriers failed to see the scaling problems coming in the core network with the explosion of Internet-driven bandwidth," says John Adler, director of marketing at Cisco. "Today, they see the need to move time-slot, packet, and cell aggregation to the edge of the optical-core network. Then, the optical backbone provides rapid provisioning and restoration for all types of traffic, consistent with a multiservice architecture."
Optical networking is being revolutionized as red-hot products are moving into lab evaluations and field deployments in record number and size. According to Adler, there is significant buzz around fully transparent systems that, for several reasons, may not prove particularly effective.
"First, some require the same DWDM vendor as the transparent switch, eliminating any best-of-breed or multivendor environment networks," says Adler. "Second, the network engineering and operations groups in the carriers have not engaged the transparent system model. These approaches are still too new and too difficult, to maintain both the element and the network."
Although most industry players are excited by the emerging all-optical network architectures, there is still enough work remaining to leave some a bit more skeptical. As with most cutting-edge technologies and products, a few vendors and carriers will "damn the torpedoes" and forge ahead. Others will take a "wait-and-see" approach before embracing and, more importantly, investing substantial capital in it.
"The hype around all-optical needs to focus on solving everyday network operation problems," say Nicholas DeVito, vice president of product development and business development at Tellium (Oceanport, NJ). "Right now, OEO [optical-electrical-optical] devices do a better job and are more scalable, reliable, and flexible. As devices for all-optical-switch cores become available, probably next year, you'll see the next generation of switch fabrics introduced."
But new entrants into the OXC market are continuing to respond to the carrier needs, introducing solution after solution. Innovation in provisioning techniques, port densities, better restoration capabilities, scalability, interoperability, and reconfigurability is providing startups with plenty of niche opportunities. However, the one critical element to success in the market-a truly proven product that provides all the benefits vendors and pundits have promised-remains elusive. This, of course, is due in part because the technology is so new.
"Continued development expenditures are needed to be competitive in this market," says Alcatel's Novak. "A prototype product or concept will not be sufficient to be truly successful in the market. Market success will be established by delivering an optical-crossconnect product that evolves to meet the overall network requirements. The OXC must be robust, scalable in size, and able to add additional functionality that will be identified as the optical layer develops."
Cisco's Adler agrees, pointing out that vendors need to offer more than a single-point solution and have a solid network-management solution to compete. Additionally, a substantial time investment is required, as well as networking expertise and capital, to successfully reach deployments in major carrier networks. The ability of the vendor's offering to meet the demand of Internet-scale, carrier-class optical networks will be the "proof in the pudding."
With the industry's optical tendencies made clear, Technology Futures' Kleinebecker believes the OXC market as we know it is probably limited. Believing OXCs will shortly be subsumed by the optical switches, his advice for remaining competitive in this market sector is, to paraphrase a well-known saying, "Go optical switching, young man!"
More detailed information on these and other products can be found in the Lightwave 2000 Worldwide Directory of Fiber-optic Communications Products and Services.