The technology that underpins today's optical networks, dense wavelength-division multiplexing (DWDM), is slowly but steadily migrating from network cores toward the end user, driven by inexhaustible demands for bandwidth. Along the way, DWDM is finding a place in national and regional backbones, metropolitan optical networks, and in limited cases, the access network.
DWDM's role at the network core is straightforward: to provide as much capacity as possible by supporting multiple wavelengths of Synchronous Optical Network/Synchronous Digital Hierarchy (SONET/SDH) OC-48 (2.5 Gbits/sec) and OC-192 (10 Gbits/sec) on a single fiber. Improving the technology means increasing the number of wavelengths per fiber, distance between optical amplifiers, and capacity per wavelength. Already demonstrated, terabit DWDM transport networks will become commonplace after the year 2000.
There is almost universal agreement among backbone network providers that DWDM is the ultimate solution for raw capacity expansion. This outlook has resulted in a very robust, global optical-networking market of more than $2.8 billion for this year, increasing 20% to 23% per year to $6.2 billion in 2003 (see Fig. 1).
Uniform in their network requirements, incumbent local-exchange carriers (ILECs) and competitive local-exchange carriers (CLECs) find they must adopt a plan that allows them to maximize revenues and maintain customer loyalty during a technology transition period. LECs' networks are evolving from time-division multiplexing (TDM) traffic to the "new public network," in which the vast majority of services will be carried on a common Internet-protocol (IP) or packet-based infrastructure. The new public network will require highly scalable solutions, with virtually limitless transparent bandwidth, to drive data-rich applications.
Yet at the same time, these carriers need to deploy networks that provide the lowest cost per bit at a level of reliability equal to or greater than today's SONET/SDH-based infrastructures. Metro-DWDM vendors promise these capabilities, and carriers are waiting to see if they deliver. Major Internet service providers (ISPs) such as Cable & Wireless USA and America Online have already announced major metro-DWDM contracts, citing at least three factors behind their technology choice:
- Transparency: The technology can support SONET/SDH and "native" data-signal formats such as Asynchronous Transfer Mode (ATM), Gigabit Ethernet, enterprise systems connection (Escon), and Fibre Channel.
- Scalability: DWDM can quickly accommodate demand for capacity on point-to-point links, and more importantly, on individual spans of existing SONET/SDH rings.
- Dynamic provisioning: DWDM offers the ability to provide high-bandwidth services such as virtual-private-network and broadband private line in days instead of months.
As these three characteristics demonstrate, the driving force in the metro-DWDM market today is economics. When the economic advantages are clear, as in the case of fiber exhaust, metro-DWDM systems show rapid growth. An in-depth analysis is warranted, however, when the economic incentives for metro-DWDM deployment depend on unique factors such as lifetime costs, network simplification, or network-element displacement.
All of the metro-DWDM deployments to date have had clear economic advantages; it is cheaper than deploying new fiber or less expensive than a "forklift" upgrade of an existing SONET/SDH ring. For the market to flourish, metro-DWDM systems must prove advantageous in deployments where the economic benefits are less clear, for example, in replacing existing network elements, consolidating overlay networks into a unified architecture, or obviating the need for additional network elements to provide new services.
Carriers today must face the daunting task of choosing among a range of unproven technologies that promise to revolutionize existing network infrastructures and services. Emerging multiservice access equipment can perform aggregation on the customer premises, in LEC access nodes, or within a central office (CO). Multiservice switches and routers may also provide transport from the same network element on protected ATM virtual path (VP) rings.
The transport choices for carriers between the CO and the network edge include SONET/SDH, ATM, WDM, and hybrids of these three. Between the network edge and the customer's premises, bandwidth and quality-of-service demands vary widely among customers. Carriers must choose among TDM, ATM, digital subscriber line (DSL), broadband wireless, and emerging optical-access technologies to address the multiple market segments in this "last mile" most effectively.
Determining which combination is most cost-effective depends on the type of service and predicted growth patterns within the metro area. For emerging carriers, the lowest-cost solution is likely to win in the near term, as long as the vendor offers a migration path. For incumbent carriers, a system that provides high levels of integration with existing network elements is likely to succeed, although this may require vendors to provide equipment that supports "pay-as-you-grow" scalability.
In Table 1, the choices available to a metro carrier today are illustrated with the associated advantages and disadvantages. The findings in the "most cost-effective access service" column are based on Pioneer Consulting's assessment of the network and operations costs associated with each solution.
From this preliminary analysis, available metro-DWDM solutions prove cost-effective only when services equal to or above OC-12 (622 Mbits/sec) are provided to customers. In the early stages of the market, this capability will drive metro DWDM to the core of these networks, where aggregation and concentration of client signals have occurred within the customer's premises or with access nodes. DWDM serves as a flexible, scalable, fiber network core, reducing the amount of fibers necessary to support service expansion, while providing transparent optical interfaces to the network edge.
The challenge to vendors is to develop cost-effective, fiber-based access solutions at rates below OC-12. This development will open up the largely untapped small and medium enterprise (SME) market. A "disruptive" new product class that combines the scalability of DWDM, the manageability of ATM, and the low cost of DSL, will lay the foundation for a lasting optical-access market, serving both large and small enterprises worldwide.
The metro optical-networking equipment market has evolved over the past two years as vendors react to carriers' requirements for higher-capacity multiprotocol infrastructures in the core and access areas of their networks. As products debut, at least three distinct segments are emerging within the metro-DWDM market.
Metro core: This area of the network includes connections between LECs' central offices (interoffice facilities) or major ISPs' or interexchange carriers' points-of-presence (PoPs). It is considered the core because the connections are between carriers' PoPs; they do not directly interface with end users. The bandwidth requirements in this segment are rising rapidly, and metro DWDM offers a cost-effective alternative to scaling these networks by adding additional fibers and stacking rings.
Metro DWDM is particularly useful in the core because it allows for network integration, in which a mix of traditional TDM and emerging packet traffic can operate over a common infrastructure, rather than overlay networks. This arrangement simplifies the core and allows for a more graceful migration from the TDM-paradigm to an IP-based network.
Products suitable to this segment are high-channel-count systems that allow rapid provisioning of high-bandwidth circuits. This market is emerging quickly as significant LECs find their fiber-optic routes exhausted and major ISPs require high-bandwidth circuits in days rather than months.
Metro access: The metro-access market is much more complicated; it consists of the rather nebulous segment between carriers' PoPs and access facilities. Access facilities may reside at an individual user's premises, a basement of a multi dwelling or multitenant unit, or an access concentration point near a cluster of end users. These facilities house access equipment, including SONET/ SDH add/drop multiplexers (ADMs), ATM service access multiplexers (SAMs), IP and public-switched telephone network (PSTN) mediation devices, and DSL access multiplexers (DSLAMs). The metro-access market is primarily served today by SONET/SDH ADMs, ATM switches and routers, and a growing number of multiservice access multiplexers.
The application of DWDM in the access segment represents a long-term strategy as bandwidth requirements exceed OC-12. In the near term, carriers can expect a plethora of solutions from multiservice access suppliers. Recent developments such as CIENA's acquisition of Omnia, the Express GX line announcement by Nortel Networks, and Atmosphere Networks' integrated SONET/SDH-ATM solution, indicate a rather uniform acceptance of the need for a service-layer infrastructure in this segment, instead of a simple "wavelength-to-the-customer" solution.
Long term, however, a metro-access network based on DWDM probably will be necessary in many implementations. SAMs will adopt integrated, passive DWDM interfaces to connect directly to metro-core DWDM systems, each on a dedicated wavelength channel. This setup will allow carriers to scale these networks on a node-by-node basis. The metro-access solution proposed by Chromatis Networks, for example, scales from an OC-12 ATM ring using low-cost 1310-nm optics to a multichannel ring using 1550-nm optics, once bandwidth requirements exceed OC-12 aggregate capacity. Each node is scalable with the addition of an optical-channel card.
Enterprise DWDM: This segment includes multichannel systems used to create high-bandwidth multiprotocol links between sites within an enterprise, which may include major corporations, educational and research institutions, medical facilities, and Internet and e-commerce sites. Today, enterprise connectivity is established through owned or leased fiber or through leased cap acity from a local car rier. As bandwidth demands increase within the enterprise-particularly at locations that house major Web or data servers-storage-area networks are being implemented using Fibre Channel and other high-speed interconnect protocols. Taking those networks out of the building and across a campus or metro area is possible using a low-cost, enterprise-DWDM system. IBM and ADVA Optical Networking are early leaders in this market, but other suppliers are beginning to understand the opportunities and will likely introduce products by the end of this year.
Table 2 describes each of the three metro segments and their respective applications. The metro optical-access market will comprise sales of systems that provide connectivity between COs, hubs, or PoPs and customer sites or aggregation points. This market is sometimes difficult to define because it is addressed by both fiber and copper access equipment. DSL-based solutions have emerged as the popular choice for the SME market this year, but Pioneer anticipates these services will not provide adequate bandwidth to support many business applications beyond 2003. Fiber-based solutions offer a high degree of capacity and scalability but are currently quite expensive at $30,000 to $100,000 per node.
As it stands today, the metro optical-access market is limited, not by demand, but by the expense and complexity of the available equipment. Based on today's product offerings, Pioneer has forecast total growth in the metro-DWDM market in North America from less than $100 million this year to $923 million in 2003. Of that, optical access will represent approximately one-third of the total, reaching $303 million in 2003 (see Fig. 2).
A metro optical-access network, as defined in these forecasts, consists the following basic features:
- Transport of client traffic from access realm to the network core or hub.
- Each client or access node supports a one or two wavelength add/drop.
- Optical channels can carry rates from 155 Mbits/sec to 1.25 Gbits/sec.
- Metro-access rings usually have hubs located at a CO or PoP facility and are managed from these hubs.
- Access rings rarely exceed 50 km.
- Future implementations will include dynamic ADM, optical performance monitoring, optical-layer and data-layer interworking, and higher channel counts.
As new metro fiber networks are deployed, it is safe to assume that these infrastructures will be built in a series of rings emanating out from a core. The hub of the metro-access ring, therefore, will usually be located at a CO, with two to six access nodes collecting traffic either directly from customer-premises equipment or from other access nodes that collect multiple-user data streams.
The metro-access market today is served by a variety of TDM, ATM, and integrated SONET-ATM devices, making it highly complex and competitive. For metro optical-access systems to succeed in this area, they must be both cost- and feature-competitive. Today, few systems are immediately competitive with SONET or ATM alternatives, making the near-term opportunity in this market more realistic for SONET- and ATM-based systems. As these systems benefit from integration and improved data-traffic handling, they will remain competitive with DWDM-based optical-access systems for the entire forecast period.
That said, it is likely the metro optical-access market will not represent a significant opportunity for vendors until after 2001. Some vendors, such as Chromatis and New Access, have developed hybrid solutions that combine DWDM with an ATM- or IP-based access platform, while others are looking for lower-cost methods of placing client traffic directly on the optical layer. Other systems, like JDS Uniphase's WaveShifter iWDM, use low-cost optics and widely separated WDM channels to create an access node more cost-effective for access implementations. Vendors like Sycamore Networks and Optical Networks have emphasized software in their product offerings, providing a platform for interworking between the optical and data network layers to speed wavelength service provisioning for carriers.
At this early stage of the market's development, no one solution has emerged superior to any other. Indeed, carriers remain in an evaluation phase for all metro-access equipment at this time and are busy considering what applications and services can be supported over this new infrastructure.
The market for metro optical-access networks will hinge upon the demand for low-cost access to metropolitan-area-network hubs. These hubs include metro COs, as well as ISP, CLEC, and interexchange carrier PoPs, which number more than 10,000 in the United States today. It is quite likely that the optical-access market will diversify over the coming five years. Optical-access nodes may evolve from simple optical ADMs to integrated access devices, which include service-layer interfaces, an ATM or IP backplane, and an optical transport layer. This evolution would increase the functionality and complexity of each node but would also potentially reduce the complexity of the overall access network.
The optical-access market has ample room and opportunity to expand, diversify, and improve upon the limited equipment available today. If vendors can meet this challenge, fiber-to-the-business can become as commonplace as private lines today, eliminating the bottlenecks at the last mile, while fueling the core optical-networking market, which provides the floodgates for this new rush of bandwidth.
Scott Clavenna is the principal analyst at market-research firm Pioneer Consulting LLC (Cambridge, MA).