Given the massive telecom recession and the metropolitan fiber overbuild, the last thing any pundit might have expected was a move to make fiber deployment cheaper in the last mile to home and office, and in the feeder rings leading to metropolitan Sonet rings. The cost of single-mode fiber itself, as well as the craft costs of fiber splicing and installation, have not experienced the kinds of drops that could make fiber to the curb or desktop ubiquitous.
Yet fiber creep may well be the big story at this week's Supercomm show in Atlanta. On the physical layer, new connector and transceiver concepts are making single-mode fiber links nearly as cost-effective as multimode fiber. The Optical Internetworking Forum's work in defining very short-reach optics for router and storage-area network (SAN) applications within the enterprise is being followed up with more work to reduce electrical-interface costs.
Two alternative enterprise access methods, passive optical networks (PONs) and coarse wave-division multiplexing (CWDM), are finding new interest from managers of specialized enterprise networks and of multitenant buildings. Where it was difficult to justify PONs for the residential full-service access networks of the past, the all-passive electronics used in the fiber fan-outs may be ideal for remote business parks where active optics cost too much.
CWDM, one of the original implementations for 10-Gbit Ethernet at the physical-connector level, is being touted at the system level as the most viable way to use fiber for server clusters and SANs. The manufacturers of integrated access devices are experimenting with both CWDM and DWDM line cards, combining managed time-division multiplexed channels, managed Internet Protocol flows and managed wavelengths into single multiservice switching systems that aggregate traffic on several layers at once.
At the data-link layer, two trends are helping to make fiber more affordable: Ethernet is becoming the single viable data-framing standard for both LAN and WAN, while Sonet is gaining a midlife kicker through the use of new data-framing standards such as Generic Framing Procedure and virtual concatenation. Marrying Ethernet and Sonet resembles the old candy commercial about putting chocolate and peanut butter together-and the results might turn out to be as tasty.
This will be evident at two places on the Supercomm floor. At individual vendor booths, TDM specialists like Larscom and Anda Networks, as well as resilient packet ring specialists such as Luminous and Corrigent, all will show new systems that combine Sonet and Ethernet to make fiber a more cost-effective option.
Industry coalitions also will be working together to an unprecedented extent at Supercomm. In a special "SuperDemo" area, demonstration kiosks for the Optical Internetworking Forum, Metropolitan Ethernet Forum and Ethernet in the First Mile Alliance all will show how the Sonet-Ethernet combination of service helps make fiber deployment more feasible. Even the demonstrations of the combined MPLS (multiprotocol label-switching) Forum and Frame Relay Forum will show how common Layer 2 packet flows enable aggregation over fiber.
Of course, the SuperDemos do not benefit fiber media alone. Digital subscriber line proponents will be present at the MEF and EFMA booths, showing how a turn from ATM to Ethernet helps make very high-bit-rate DSL and G.SHDSL (the new standard for symmetric HDSL) over copper media more affordable. But even the promotion of new DSL techniques from the likes of Infineon, Ikanos and Metalink has a hidden fiber message: If VDSL becomes popular for apartment buildings, it pushes fiber farther out into the neighborhood, making the case for broadband services that can play to the mutual advantage of OIF, MEF, EFMA and MPLS/FR forums alike.
The funny thing about the fiber revolution in the midst of a recession is that it required neither new infrastructure nor huge fiber price drops. Instead, this coup took the best of legacy Ethernet and Sonet systems to make aggregation of traffic over fiber more feasible-and last-mile fiber more cost-effective.
This week's In Focus has some diverse opinions from engineers at companies such as AT&T Labs, Nortel Networks and Hatteras Networks-as well as online exclusives from Adva Optical Networking and OnFiber Communications-who will provide a taste of how widespread this revolution may turn out to be.
AT&T Labs' researchers offer a perspective on the reality of a massive rollout for Fiber-to-the-Home and discuss the cost, legislative and technical obstacles that have interrupted the original vision-going back almost two decades.
Nortel Networks' manager, Rob Keates, makes a case regarding the number of business and technology developments that are directed at making "fiberization" of the last mile more compelling for enterprise customers-including making Ethernet the LAN protocol of choice, and developing storage-connectivity requirements and innovations (GFP, VCAT) to make the core network data-friendly.
Also eyeing fiber conservation in last-mile Ethernet deployment, Mark Ventura, vice president of hardware engineering at Hatteras Networks, offers a cost-cutting option-single-fiber, bi-directional technology-which he sees as "fast emerging as the low-cost solution for the Fast Ethernet access market . . . the same technology can change the cost model for GigE access as well."
In his online exclusive, OnFiber Communications' Boyd Chastant, senior product manager, examines the checks and balances of Ethernet. "The engineering and architecting of a carrier-class Ethernet network add costs to the network deployment such that it approaches that of a typical private line network," he points out.
Also online, Adva's Abdul Kasim and Brian McCann discuss how Ethernet over WDM has emerged as a powerful revenue source: "Leading European and North American carriers are experiencing a return on investment of as short as six months in carrier-class optical solutions delivering low total cost of ownership."