"Running the numbers on wavelength services"
Here's an interesting article comparing the economics of SONET vis a vis DWDM solutions. Written by Lawrence Gasman, CEO of CIR, FWIW, YMMV.
telecominvestormag.com
The Business Case: The icing on the optical networking cake
Running the numbers on wavelength services
By Lawrence Gasman
Lawrence Gasman is president of Communications Industry Researchers, Charlottesville, Virginia.
Until recently, large users of telecommunications services – such as Fortune 500 companies, government organizations and telecommunications carriers – have constructed backbone networks from fiber that they have either installed themselves or have leased from the owners of fiber networks. Such facilities are expensive and hence have made little economic sense for most users, who have instead built their backbone networks from leased lines.
But in the past few years, the options have changed. The advent of dense wave division multiplexing (DWDM) and, more recently, optical switching, have transformed the economics of fiber optic communications. In traditional fiber optics, each fiber transmits information over a single wavelength. But with a wave division multiplexer at each end of the fiber optic link, multiple wavelengths can be transported, each carrying as much information as the single wavelength could in the traditional link. By the end of this year commercial DWDM systems will carry as many as 300 wavelengths on a single fiber, and systems capable of supporting over 1,000 wavelengths have been demonstrated in the labs.
The economics have been changed by the appearance of optical cross-connects that enable wavelength services to be rapidly provisioned
The DWDM revolution has obviously had a profound impact on the economics of bandwidth. If 300 10-Gbps channels can be transported over the same facilities that once would have carried just one 10-Gbps channel, then the cost of supplying bandwidth falls by orders of magnitude – and this can translate into much lower costs to end users for 10-Gbps services.
With conventional fiber optics the end user might lease a "dark fiber" and add optronics to get a 10-Gbps pipe. In the era of DWDM, an end user might lease a much less expensive wavelength instead. Such wavelength services are already available from a handful of service providers, including Storm Telecommunications in Europe and Metromedia Fiber Networks and Global Crossing in the US.
The economics of supplying high-bandwidth services over fiber optic facilities have also been changed by the appearance this year of optical cross-connects that enable wavelength services to be rapidly provisioned. While it can take weeks or even months to provision high-bandwidth services without such devices, optical cross-connects hold open a promise of wavelength services provisioned on demand. The market for such services would include "dot-coms," Applications Service Providers (ASPs), financial institutions and broadcasters, all of who demand flexible, affordable bandwidth services. For example, suppose an e-commerce company needs bandwidth to support bandwidth demand during a special sale: it can order up bandwidth on a wavelength service for the duration of the event.
Saving costs with DWDM
Although wavelength services present considerable new revenue opportunities for service providers, this opportunity has not been the main reason why service providers have been installing DWDM systems in their networks. Instead, many service providers – primarily long-haul operators – faced with crowded networks as the result of the growth of the Internet, have installed DWDM systems as an alternative to putting in new fiber and Sonet systems. DWDM offers such service providers the opportunity to eliminate Sonet add/drop MUXs, which are costly both in absolute terms and in terms of the floor space they occupy in increasingly crowded switching centers. Sonet, in fact, is running out of power on densely trafficked routes and has yet to be successfully scaled beyond 10 Gbps. Researchers are working on 40 Gbps Sonet systems (OC-768), but these may require service providers to re-cable their networks using special kinds of optical fiber.
To understand the economics of DWDM, consider the following. Suppose a service provider has a coast-to-coast OC-48 (2.5 Gbps) link that needs to be upgraded and a budget of $25 million to do the upgrade. For this sort of money the company would probably have the option of upgrading the link using a pure Sonet solution to an OC-192 (10 Gbps) link. Or it could install, say, 16 channels of DWDM instead, with each channel operating at OC-48. Put this way, the end result is a no brainer. Using the Sonet-only solution, the investment would buy an additional three OC-48 equivalents of capacity, while the DWDM solution gets an additional 15 OC-48s.
This is just an example. The actual costs of installing new networks vary hugely on a case-by-case basis. In some cases – especially for short-haul or metropolitan networks – installing new fiber and Sonet gear may still make more sense. And Sonet comes equipped with much more sophisticated network management capabilities than optical networking gear, which makes it especially attractive to the established service providers, who tend to be quite conservative about such matters.
However, while the actual numbers may vary from company to company, there is no doubt that optical networking is bringing significant operational cost advantages to service providers. In 2000, US service providers will spend about $3.8 billion on optical networking gear to obtain these benefits. Wavelength services are just the icing on the cake. And at present they are a fairly thin layer of icing on a rather large cake. Most users of wavelength services are other service providers of various kinds, but large end users are also potential subscribers to such services. For example, Freddie Mac, the mortgage corporation chartered by Congress, is a subscriber to Metromedia Fiber Network’s wavelength service.
For the time being, wavelength services would be of interest to only such very large users. This is because it makes little engineering or economic sense to provision services at speeds of less than OC-48 on a single wavelength, and only the largest businesses and government organizations could possibly want this much bandwidth. For these users, however, the economies implicit in optical networking, and which are discussed above, can result in lower prices than other options – these other options being primarily leasing dark fiber or leased lines.
In addition, optical cross-connect facilities will potentially enable wavelength services to be deployed in minutes, rather than the weeks or months required for high-bandwidth leased lines. These optical cross-connect facilities potentially can offer a level of network management sophistication that is on a par with what is available for leased lines, and superior to what is available for dark fiber.
Wavelength services are not widely available at the moment, but they will probably have some impact on the high-end leased line and the dark fiber market in a few years. However, these markets are themselves growing so fast that this impact may be small. Whether wavelength services will ever reach down to the smaller user is questionable. But new ways of bringing the virtues of optical networking all the way to downtown buildings – technologies such as passive optical networks and optical wireless – coupled with growing bandwidth demands from all sizes of user, may ultimately make this possible.
The wavelength services business case
So how will carriers actually make a business case for wavelength services? This will depend very much on their internal needs for bandwidth, their opportunities for selling wavelength services and whether they will be winning new customers for wavelength services, as opposed to transitioning customers from leased lines or dark fiber services. An example of how a business case may be made is set out in the accompanying chart, which takes the example given above as a starting point.
Let us suppose for the sake of argument that the service provider in question needed 15 OC-48 equivalents, of which he was going to use 10 OC-48s internally and sell off five OC-48s. The annual revenue that would result would vary considerably depending mostly on distance, but a reasonable guess would be $5 million per year per OC-48. Thus the five OC-48s in our example would produce revenues of $25 million per year.
From this figure must be deducted any revenues that might be displaced as existing customers move to the newly available bandwidth. In some cases, the value of displacement would be zero, where a start-up service provider is winning entirely new business. At the other end of the scale, an established service provider might be transferring an existing dark fiber customer to wavelength services, but may be gaining some additional revenues from managed network services associated with the wavelength services. In many cases, there will be some lost revenues as a customer ramps up from some relatively modest leased line usage to big wavelength-based pipes. For the sake of our example, we show that 60% of the revenues from wavelength services go toward displacing existing revenues.
One would also have to factor into the equation the annual cost savings from deploying the optical networking solution rather than an equivalent Sonet solution, remembering that to get the additional capacity of 15 OC-48s using a Sonet solution would mean an investment of $100 million or more. The $25 million only bought an additional three OC-48s. Also, operational costs, including depreciation, for the Sonet approach could cost as much as $40 million, while we estimate the equivalent costs for an optical approach at around $9 million.
As the accompanying chart shows, such assumptions produce a payback of slightly over seven months, which makes buying the optical networking solution another no brainer. It is important, though, to realize just how case-sensitive this analysis is. A service provider that had no potential customers for wavelength services and needed only modest upgrades of its capacity may still see Sonet as the way to go. |
