Introduction

From its earliest days in the 1980s when fiber optics was first being deployed en masse right up to the present day when DWDM and EDFAs flourish, choosing optical networking over a copper (or wireless) solution has been a trade off between bandwidth and systems cost. If you choose fiber, you get a lot more bandwidth, and the cost per bit may be less than alternative media. But in absolute terms, you also pay more for photonic hardware than you do for copper or wireless hardware.

Because of this disparity, vendors and R&D facilities have long sought ways to reduce the cost of optical networking. For reasons that are explained below, Communications Industry Researchers (CIR) believes that this task has never been more urgent. This paper discusses some innovative ways that are now being proposed for making optical networking less expensive. Much of the information in this paper is taken from two research studies recently published by CIR – "WAVE DIVISION MULTIPLEXING, PHOTONIC SWITCHING AND THE COMING OF ALL OPTICAL NETWORKS" and "FIBER-TO-THE-X: THE NEW MARKET FOR FIBER TO THE DESK, CABINET AND HOME."

While bringing down the cost of optical networking may have always been an "issue," CIR believes it is now an imperative. If equipment vendors and component manufacturers cannot lower the cost of optical networking, CIR believes that they will be unable to exploit four major opportunities that are now emerging. These are:

• The bulging metro environment. The metro segment of the public network is a growing bottleneck as end users jack into the Internet at broadband speeds using low cost methods — xDSL and cable modem access. Terabit backbones are being constructed using DWDM technology and it would be natural to transfer this technology to the metro space. However, in most locations DWDM is still too expensive. CIR believes that by 2003, U.S. ILECs and CLECs could be spending as much as $1.9 billion on DWDM systems, but only if the cost of such systems can be brought down in price by 30 to 40 percent.

• The residential fiber opportunity. Fiber-to-the-home has been a futurological fantasy for almost two decades. CIR believes that it will stay that way for some time. But fiber-close-to-the-residence (a.k.a, fiber-to-the-curb or fiber-to-the-cabinet) is very much a market opportunity. While xDSL technology gold plates the telco network, its distance limitations, bandwidth limitations and mixed throughput performance will ultimately doom it to being an interim technology. Major telcos, such as BellSouth and SBC, have made commitments to bringing fiber much closer to the residential customer than ever before -- although the network closest to the customer will remain copper. Such projects implicitly assume that the cost of close-in fiber will decline significantly. Assuming that the necessary cost improvement can be achieved, CIR believes that the 2003 market for fiber-to-or-near-the-home could be as much as $687.5 million.

• Fiber-to-and-in-the-business. Most large commercial and industrial businesses are now served by fiber-to-the-building and the trend is also beginning to impact smaller buildings. According to CIR estimates, almost 100,000 buildings in the U.S. are served by fiber-to-the-building and by 2003 this number will have reached 136,000 buildings. In addition, the inside of buildings are also getting more photonic. CIR believes that the fiber-to-the-desk market will burgeon from about $108 million in 2000 to $435 million in 2003. Once there was little need for fiber-in-the-building. The mainstream use of Gigabit Ethernet and Fibre Channel has brought fiber-to-the-wiring closet in many buildings. Soon it will bring fiber-to-the-desk. Canoga Perkins, ADVA, Inrange Systems and others are already doing a brisk business in DWDM systems for enterprise applications. But for fiber-to-the-wiring closet and fiber-to-the-desk to be useful to smaller businesses, the cost of fiber in-building systems will have to fall.

• The flexible bandwidth issue. There is a growing need for optical switches and crossconnects to provide bandwidth-on-demand for special events on the Web, disaster recovery, etc. Such switches are already beginning to appear and are being used by "next generation" service providers such as Qwest, Enron and Williams. However, the same functionality will be required throughout the network and 3M has already announced that it is working on a premises-based optical switch. Whether such products will ever find a market will depend strongly on whether they can be made cost competitive with electronically based LAN switches. CIR will soon be publishing a new study analyzing the opportunities available in the optical switching market over the next few years.

Bringing down the cost of optical networking is an area that has been of much concern to the telecommunications industry since the days that the first fiber optics were being installed between switches. Many different approaches have been suggested. However, when analyzed, CIR believes that they all fall into a relatively small number of strategic categories:

• Mass Production and Experience Curve Effects. The hope is that as end users – be they carriers or enterprise networks – begin to take advantage of fiber optics, enough optical networking components and systems will be sold to reduce the cost of optical networking, and manufacturing processes will improve. This strategy has worked extremely well in the past – for example, the cost of single-mode fiber has been brought down to the same level as multimode.

• Architectural Improvements. The productivity gains that FSAN expects to be translated into lower costs for close-in fiber are obviously predicated on standardizing an architecture – in this case PONs (passive optical networks). However, the architecture chosen for close-in fiber can impact costs in more ways than through standardized components. Some vendors are recognizing that a single architecture must today support users that need the power of optical networking and those that do not yet have this need. An example here is Wavesplitter Technologies, whose WaveXpander product claims to be the first and only DWDM expansion architecture to offer a modular pay-as-you-grow upgrade path to increased DWDM channel counts. This means that service providers need pay only for the incremental costs for higher channel counts on an as-needed basis. A similar value proposition is being put forward by Chromatis Networks, whose Selective Wave Division Multiplexing (SWDM) technology enables service providers to deliver one or more wavelengths to those sites that require multi-gigabit capacities, while allowing other sites on the same ring to share less-expensive bandwidth. Chromatis’ value proposition recognizes that while most businesses today can benefit from the bandwidth that fiber brings, they may vary considerably in the amount of bandwidth they require.

• Using Less Expensive Media and Components. Redesign of architectures is not the only design issue that needs to be considered in cost reduction for close-in fiber. The redesign of fiber optic products to reduce costs may also be key. This has been proven with the arrival of Small Form Factor (SFF) connectors that cost less and have a smaller footprint than the earlier generation of fiber optic connectors. This smaller footprint also leads to cost reductions – for example, through enabling more ports on a card. Another way to lower the costs of close-in optical networking is to use lower-cost/lower-performance components. For example, LEDs are used instead of lasers. Or multimode (or even plastic) fiber is used instead of single mode. This approach can be somewhat controversial, since fiber optic purists claim that leading edge technology should always be deployed – even close to the customer – in order to futureproof the network. Thus, the backers of FSAN and many proponents of Gigabit Ethernet say single mode is the way to go.

• Making Low Cost Components Have Higher Performance. One way round the future proofing problem is to make lower-cost/lower-performance components work better. This is a strategy that has worked extremely well in the copper environment. Copper cabling has been improved to support Gigabit speeds over short distances, for example.

The Exhibit below profiles the four strategies that are discussed above for lowering the cost of optical networking.

The reader should also note that the companies that are used above to exemplify the four strategies are by no means the only innovative new companies that are coming up with new ways of making optical networking closer and cheaper. Other companies that deserve recognition in this regard include Quantum Bridge Communications — which also wants to bring the virtues of optical networking to almost any size business, and is using PONs to do this. Also included is the NEC subsidiary, Eluminant, which has a product called Sequel which is targeting the business that requires more than DSL capability but that is not ready for that leased OC-3.

It seems likely that there will be growing attention paid to close-in optical networking during the course of the year 2000 and beyond. As we indicated in the first part of this paper, there is plenty of potential for new business revenues locked up in this market – but only if the costs can be brought down.

************************

This White Paper is published by Communications Industry Researchers, Inc., a Charlottesville, VA-based market research, forecasting and strategic planning firm focussing on new opportunities in optical networking. For more details, please contact Rob Nolan at 617-923-7611 or ROB@CIR-INC.COM. More information about CIR can also be obtained at www.cir-inc.com.