dwdm heralds the emerging multiwave optical layer
Dense wavelength-division multiplexing technology will provide the flexibility to meet future bandwidth needs and will change the way sonet is employed.
Stephen B. Alexander and Joe Berthold
ciena Corp.
Spurred by the growth of demand for bandwidth--especially from data services such as the Internet--operators are studying ways to institute fundamental changes in their network architectures. At the core of this thinking is how they can take full advantage of dense wavelength-division multiplexing (dwdm), which enables carriers to send multiple wavelengths across a single strand of optical fiber. (See "How dwdm works" on page 43 for a tutorial on the technology.) We believe the emerging multiwave optical layer will have a profound impact on network architectures because it offers a simpler and less expensive way to carry high-speed data.
Fiber-optic networks began carrying traffic approximately 20 years ago. There was a much-heralded evolution to Synchronous Optical Network/Synchronous Digital Hierarchy (sonet/sdh), which has provided a strong interface standard that promotes interoperability, protection mechanisms, network management, and time multiplexing hierarchy.
Existing network architectures deploy sonet/sdh as the transport system, possibly working in collaboration with dwdm. A number of services are carried over sonet transport systems, including voice, frame relay and other data services, atm-based services, and the Internet (see Fig. 1).
Changing the network paradigm
Yet a major problem with the equipment in today`s networks is that it was designed to manage traffic in the basic unit of a telephone call--64 kbits/sec--while future traffic growth will be mainly in cell- and packet-based services. Data services today are often carried over T1 circuits at 1.554 Mbits/sec or, for high-end users, T3 circuits at 44.736 Mbits/sec. But this bandwidth is increasingly not enough to satisfy the needs of users, who are able to consume sonet OC-3 (155-Mbit/sec) and OC-12 (622-Mbit/sec) channels. When carriers look to their options for provisioning such high-bandwidth services, they find that the equipment at their disposal is more complicated and expensive than they need, since it was designed to manipulate much more fine-grained traffic than is currently required.
Data networking equipment such as Asynchronous Transfer Mode (atm) or Internet protocol (IP) switches and routers can handle data traffic much more efficiently than do telephone voice switches. Data equipment will become the leading source of traffic for transport networks. Data switches will manage bandwidth and aggregate traffic; the transport network will provide low-cost and reliable connections between them. About one-half of the traffic in the network today is data, but the majority is carried on low-speed channels in voice switches or sonet multiplexers. Data growth is dramatic and is estimated by some sources to be 35% annually (see Lightwave, May 1997, page 41). In 1998, data networking equipment will incorporate interfaces at 2.5 Gbits/sec and will begin to be introduced into Internet backbone networks.
The performance of data equipment, measured in port speed and total traffic throughput, is rapidly increasing. Early next year, data equipment will have output ports that run at the speed of fiber-optic transmission systems, most commonly 2.5 Gbits/sec. These switches will thus have done all the multiplexing necessary for transmission, leaving further levels of sonet multiplexing unnecessary. sonet will continue to be the premier interface standard for wide-area transmission equipment. However, the variety of equipment that embodies the sonet standard is broadening to include data and dwdm networking equipment.
The real message of dwdm
In many respects, the further implementation of dwdm in the network simply verifies that network operators believe dwdm can help them get more out of an optical fiber than other techniques. Optical fiber can potentially transmit trillions of bits of information per second. It makes sense to center the network around a transport technology that enables operators to more effectively and efficiently use that potential. dwdm is increasingly employed when the traffic exceeds the limits of what is economical to transport with time-division multiplexing.
dwdm offers another advantage to operators in these turbulent times: It gives them a tool to help deal with uncertainty. Competition and the explosion of new data service demands put carriers in a quandary. Should they install new fiber and transport equipment so they are ready to meet any demand quickly? This could result in a large, stranded investment if either the traffic does not materialize or competitors win some of their customers. Because dwdm equipment can be designed to be modular--costs are driven by the wavelengths provisioned--installing dwdm systems defers the expense of new cable installations and still positions carriers for dramatic service growth. Since dwdm systems that use an open architecture are protocol-transparent, they are a safe investment no matter which camp--atm or IP--wins the protocol wars. All of these features of dwdm networking help a carrier conserve capital.
Network operators who believe that future traffic growth will be dominated by data now see the opportunity offered by dwdm: It can lower their costs and simplify their networks. dwdm allows them to remove an entire class of equipment in their emerging, high-capacity, data backbone networks. sonet multiplexers are no longer needed because time-division multiplexers are no longer needed. The last stage of data aggregation is done by cell and packet switches, and the bandwidth of optical fibers is then most efficiently used by combining a large number of wavelengths, each carrying a high-speed data channel.
It is easy to understand why this change has not yet occurred and why it will take time to play out. The vast majority of data equipment in backbone networks still has lower-speed interfaces such as 45 or 155 Mbits/sec. dwdm will soon fill a backbone data need and then move further toward the edges of the network. Also, as wavelength counts increase and manufacturing costs decrease, dwdm will become more economical in regional and access networks, even with channel speeds of 155 and 622 Mbits/sec.
Changing role of sonet/sdh
No one knows better than the network operator that networks are progressive, ongoing entities. However, one of the benefits of dwdm is that it can easily be overlaid with existing networks, seamlessly dwarfing the capabilities of the original network. The tremendous growth in data suggests an overlay network that can meet this demand and also allow for growth.
We believe that the current generation of sonet networking equipment, including terminal multiplexers, add/drop multiplexers, and crossconnect switches, will continue to serve the needs of low-bandwidth, circuit-oriented traffic, but that sonet equipment will no longer be the only choice for optical network transport equipment. dwdm networking equipment will provide a new set of network elements, including dwdm terminals, dwdm add/drop multiplexers, and optical channel crossconnect switches. dwdm networking equipment will fill two roles: It will economically transport high-bandwidth sonet channels, which are the outputs of sonet equipment, and will carry the whole new class of data networking equipment that will form the basis of most of the traffic growth in the future (see Fig. 2).
sonet will continue to be the interface standard of choice, because without standard transport rates and formats interoperability is impossible. Low-cost sonet interface components, in the form of the sonet 1300-nm short-reach interface, provide the most economical way to interconnect high-bandwidth channels to dwdm networking equipment, be they from sonet terminals or data equipment.
dwdm equipment will build upon the sonet standards, implementing mechanisms for channel identification and monitoring to make dwdm networks manageable. Signal regeneration will be integrated into dwdm terminals to ensure scalability, so complex metropolitan-area and national-scale networks can be assembled simply. Protection mechanisms will be introduced so data services that require a high degree of survivability can obtain it from the dwdm layer without the added expense of sonet rings. At the same time, the dwdm layer will continue to support sonet rings.
dwdm networking equipment will support the management protocols desired by large network operators--the Telecommunications Management Network standards--and those desired by data network operators and end-users--simple network management protocol.
We see sonet`s new role as one of several protocols connecting to dwdm, which in turn, transmits multiple signals through the fiber (see Fig. 3). We believe this represents a more realistic scheme that best describes how networks will evolve.
dwdm: A complete network architecture
dwdm began as a way to alleviate telecommunications traffic congestion in the long-distance marketplace. However, it has become clear that the emerging multiwave optical layer will be used throughout the telecommunications network for both long- and short-haul applications. dwdm networking will allow high-bandwidth channels to be delivered across complex networks in a simple, manageable, and cost-effective way. This simplification of some of the key pieces of equipment in wide-area networking will provide network operators with the tools necessary to offer more-economical, high-bandwidth services, which will lead to an even faster escalation in traffic.
The strategy behind the emerging multiwave optical layer is to provide a clear and seamless dwdm network in a manner that can interconnect with and allow existing networks to continue to function. Only then will network operators be able to truly future-proof their networks against the bandwidth jolts and spikes that will continue to mark our information age. u
Stephen B. Alexander is vice president of transport products and Joe Berthold is vice president of network architecture, both at ciena Corp., Linthicum, MD.