Metro DWDM: Will the dogs eat dog food?

Th 79310

The metro market has been touted as the next big play for DWDM, but how soon is 'next'?

BY PETER J. FARMER

For more than a few years, we have been told of the advantages of transparency and DWDM for metropolitan networks. Products have appeared at both the component and system level. Dreams have been spun and fortunes made, and lost, based on the allure of protocol- and bit-rate-independent networks that could deliver wavelength-level services at a fraction of today's prices.Th 79310

Peter J. Farmer is director, optical-networking practice, at Strategies Unlimited (Mountain View, CA). He can be reached at pfarmer@strategies-u.com.

The appeal of transparent DWDM networks re mains, but implementation has been slow. In the parlance of the venture-capital community, the dogs aren't eating the dog food. Let's examine this seeming lack of appetite.

The first metro DWDM systems were introduced in 1998 and '99 by the vendors of long-haul DWDM. Fundamentally, these were point-to-point systems to relieve fiber exhaust. Their designs were similar to their long-haul cousins but with smaller channel counts, fewer optical amplifiers, and lower prices.

Sales were less than brisk, for as it turned out, fiber exhaust was not as pressing an issue for metro carriers as it was for AT&T, Sprint, and MCI. The long-haul carriers were facing an explosion in traffic, and without point-to-point DWDM systems, their only option was to lay new fiber in networks that stretched for tens of thousands of miles. DWDM came as the answer to a prayer.

Distances between nodes in incumbent local-ex change carriers' (ILECs') networks were measured in miles, not hundreds of miles. Because the ILECs had replaced much of the fat copper wire that had once filled their underground conduits with skinny fiber, they often had conduit runs available for pulling new fiber. While fiber exhaust was occasionally an issue for the ILECs, that was the exception rather than the rule.

Nor did the upstarts-competitive local-exchange carriers (CLECs) and the builders of metropolitan dark fiber networks-have much interest in point-to-point systems. Dark fiber builders were tearing up streets to lay bundles containing hundreds of fibers. CLECs, for the most part, were planning for the future and looking to second-generation metro DWDM systems to deliver services over these networks.

Those second-generation systems approached DWDM quite differently. Prominent in the second-generation systems, which emerged in 2000, was the optical add/drop multiplexer. The second-generation systems were to be implemented as part of a new network architecture. DWDM delivered capacity on these new networks; optical add/drops provided the means to deliver that capacity to the end user. Channels that did not have to be dropped at a particular location could be "expressed"-sent through a network node without optical-electrical-optical conversion, thereby reducing infrastructure costs. Networks would be able to provide services based on non-SONET protocols such as Gigabit Ethernet, Fibre Channel, and Escon as well as wavelength-level services. Additionally, the new equipment and new architectures would enable faster provisioning of service, eliminating many of the time-consuming, error-prone manual steps needed in traditional ILEC networks.

These efforts have not been without success. Nortel Networks, Ciena, ONI Systems, ADVA Optical Net working, and others have sold next-generation metro DWDM systems-principally to newer carriers-and benefits from their new approaches are being realized. Nevertheless, this area is characterized more by disappointment than elation. Why has the uptake not been faster?

The key reason for less-than-hoped-for results is that the competitive picture changed. With few exceptions, the startups developing metro DWDM systems were banking on purchases by the CLECs. These newer carriers, they reasoned, would be more receptive to new ideas and could act much faster than the ILEC "dinosaurs."

Unfortunately, a number of the CLECs ran out of money before purchase orders materialized. For metro DWDM systems to generate significant revenue, they would have to be sold to the ILECs.

Implementation of these new systems in an ILEC network is much more problematic than in a "greenfield" CLEC network. The ILECs have a vast base of legacy SONET equipment at 1310 nm, implemented and managed via an alphabet soup of order-entry, asset-management, and billing systems. These operational support systems (OSSs) were born in the pre-divestiture (pre-1984) Bell System.

Not surprisingly, OSSs are less than elegant. One such system, Trunk Information Record Keeping System (TIRKS), used by ILECs to maintain an inventory of circuits, has been described as "basically one of the largest and most despised software systems in the world"-one that "makes Unix seem wicked friendly."

Ugly as they may be, these legacy OSSs are associated with operations that continue to generate prodigious levels of cash. The money machines known as ILECs produce tens of billions of free cash flow each year. ILECs are understandably reluctant to make any wholesale changes to networks that have been so successful in generating cash.

Consequently, those who wish to sell metro DWDM systems to ILECs have had to learn a new acronym: OSMINE (Operations Systems Modification of Intelligent Network Elements). The OSMINE process, run by Telcordia Technologies, involves making changes to a vendor's system hardware and software so that a particular piece of network equipment can interoperate with legacy OSS. The process is lengthy, painful, and expensive. The prospect of further delays in making sales, plus the cost of the OSMINE process itself, has led to retrenchment for many of the newer players in the metro DWDM systems market.

The prognosis for the remaining players is less than rosy. The demise of CLECs has reduced the amount of competition in local markets, leaving ILECs with the view that they are once again in control of the pace and direction of network evolution. ILECs only move quickly when they sense an acute threat to their position. DSL implementation, for example, came about only because the ILECs felt threatened by cable modems.

Moving from a one-wavelength-per-fiber, 1310-nm network to a 1550-nm DWDM network is a formidable task, and in spite of the intentions of the Telecommunications Act of 1996, the ILECs do not feel genuinely threatened by competition.

So what could get the ILECs "off the mark"? One intriguing scenario would be if the ILECs began to poach on each others' territories. Suppose that as a result of financial "restructuring," a young carrier with extensive local fiber-Level 3, for example, or Metromedia Fiber Network-were to fall into the hands of SBC Communications or Verizon Communications. That would mean facilities-based competition between companies with comparable levels of financial resources. In such a case, the "poacher," free from legacy baggage, would be inclined to install the new generation of metro DWDM equipment, thus jump-starting the market.

It is an interesting possibility but can hardly be regarded as likely. Verizon and SBC understand that poaching, like nuclear war, could lead to "mutually assured destruction."

As a result, it is difficult to be optimistic about the prospects in 2002 for metro DWDM players. We may need another year before the big dogs are ready to take a bite.

More in Network Design