Metro DWDM Rhetoric Fails to Win Contracts

By: Mark Lutkowitz

Several large telecom manufacturers were blindsided by explosive demand for long-haul dense wavelength division multiplexing (DWDM) equipment in 1996 and scrambled to ramp up manufacturing capacity. In stark contrast to the robust long-haul market, executives from metro DWDM vendors spend a great deal of time promoting the potential of the technology at conferences, rather than negotiating supply contracts with actual customers.

Given that imminent market growth would prompt vendors to keep quiet rather than publicize the opportunity for their competition, earnest conference presentations indicate that vendors see no market exuberance for metro DWDM. Despite their efforts, vendors have not convinced carriers that both the business case for metro DWDM and adequate functionality for optical networking will be available any time soon.

Looks good on paper
The optical networking intelligentsia agrees that DWDM makes sense in metropolitan public networks—at least on paper. One main DWDM application is providing optical pipes for Internet protocol (IP) routers, instead of transporting data traffic over synchronous optical network (SONET) equipment. Performance and cost objectives to make this application economical have so far eluded designers of metro DWDM products. Until an equipment supplier can meet the cost objectives and capability of an OC-48 (2.5 Gb/s) SONET ring on a DWDM system, carriers will not adopt DWDM and IP networking in lieu of SONET in this space.

The US market for DWDM in metro public networks reached not much beyond $30 million in 1998, and the figure will likely come in well under $100 million for 1999. For the next five or six years, the world market for long-haul DWDM—which may exceed $3 billion just for terrestrial applications—will probably continue to dwarf that of the short-haul metro market. The metro DWDM market will be lucky to reach one billion dollars during this period.

Different economics for different networks
The very big differences between the long distance and local networks indicate why long-haul carriers have gobbled up DWDM systems with abandon, yet local carriers generate much slower sales cycles. Cohesive long-distance networks lend themselves well to cost-effective DWDM deployment, while a patchwork of metro network operators and applications hinder across-the-board deployment of DWDM in local networks.

In long haul networks, DWDM applications generally reside within the same carrier network. Short-distance metro applications link carriers to carriers, carriers to end-users, or end-users to end-users. Unlike long-haul networks, metro networks host a very rich set of applications served by more than one kind of product. Local exchange carriers design and build metro networks almost one application at a time, determining the attractiveness of a particular solution on a case-by-case basis. Interexchange carriers design long-haul networks as a whole, which enables cost-effective DWDM deployment throughout the entire network construction.

Tough metro characteristics
The piecemeal adoption of equipment solutions in the metro network contributes to prohibitive metro DWDM cost because local exchange carriers will not buy DWDM gear in nearly as high volume as interexchange carriers have—especially in the near future. Although some manufacturers have lowered equipment cost by designing metro DWDM products from scratch instead of adapting long-haul technology to short haul applications, they have a hard time overcoming system cost barriers, such as producing less expensive, short-haul OC-48 transponders.

The metro network's manageability needs also make DWDM applications more challenging than long-haul deployments. Because metro networks are close to the end user—the source of revenue—additional management capability is even more important with metro DWDM equipment to ensure an adequate quality of service demanded by subscribers.

Optical networking
All-optical networking, replete with management intelligence, presents an attractive proposition for metro carriers—but the capability is a long way from commercial fruition. To become a bona fide network play on a grand scale in the metro environment, a DWDM system must handle thousands, perhaps tens or even hundreds of thousands of wavelengths.

Without a critical mass of wavelengths, carriers cannot justify the expense of the new network elements and operations required in the shift to a new networking paradigm. Given only a few dozen wavelengths in a portion of a network, metro DWDM only brings capacity gain to the network equation, which does not comprise a compelling value proposition for carriers and limits the technology's market potential.

Today's optical ring architectures in the public metro DWDM space perform over limited span lengths and face tough design challenges. Each node on an add/drop ring contributes to signal loss, and some platforms introduce loss with each additional channel. Loss necessitates optical amplifiers, which introduce cost, complexity, and equipment failure risk into the system, which ruins the economic proposition of the ring architecture. Manufacturers are developing low-performance amplifiers to overcome loss at low cost, but it may take them a while to produce products in high volume cost-effectively.

Commercially viable technology for all-optical networking is a long way off. Volume production of totally-optical cross-connects with switching fabrics that scale to hundreds of wavelengths may be at least six to seven years away. Programmable optical add/drop multiplexers, another important prerequisite to optical networking, will not see volume deployment for at least two years. Given that the metro marketplace changes rapidly, it is essential to deploy a network element that can dynamically rearrange, add, and drop wavelengths for an optical network. The challenge may not be so much to produce such a device, but to deliver it cost effectively.

Alternatives
Metro network operators also have attractive alternatives to DWDM. Carriers that have tapped out capacity may find it easier to add more fibers than deploy DWDM on existing fiber. Upgrading the time division multiplexing (TDM) rate may also prove more cost-effective than installing DWDM.

Speaking of TDM, improvements in SONET, the incumbent metro fiber network platform, may preclude the need for DWDM in many cases. SONET systems, designed for voice traffic, do not handle data communications effectively, yet data dominates new capacity needs. But new SONET systems such as those from Cerent (Petaluma, CA) pull together multiple types of data traffic and transport it in its native form within a SONET ring. Other suppliers have established an interoperability group, Data Aware Transport Activity (D.A.T.A.), to address data-over-SONET transport. As the rigidity of SONET systems lessens, the advantage of DWDM's protocol transparency diminishes.

Given the nebulous state of the public metro network equipment market, suppliers have not been busy signing supply contracts for metro DWDM equipment. The vendors' circular reasoning posits that if they talk about metro DWDM opportunities enough, customers will eventually get fired up enough to buy the technology. Manufacturers have been pushing metro DWDM technology in this way for more than two years. But carriers, for the most part, have not been convinced that metro DWDM offers something they do not already possess.