Different window, same view

Stephen M. Hardy

Editor in Chief

This month`s Special Report section on "Gigabit Data Networks and Equipment" opens with an article on the prospects for Fibre Channel and how application of the protocol-independent data transport standard might affect the use of fiber-optic technology. I want to take a few moments here to underscore what I discovered about the opportunities the standard might open for fiber optics--and the rather familiar hurdles that our community must overcome if we are to take advantage of those opportunities.

I won`t be giving away the end of my story to tell you now that one of Fibre Channel`s most promising applications lies in combining storage devices and servers into their own enterprise networks. Those in the know call the application "network attached storage." The premise is that as data availability and management become increasingly critical to the success of businesses of all shapes and sizes (data storage now represents 50% of the typical information technology manager`s budget in many cases, one source told me), information technology managers will soon see the benefit of distributing storage around their enterprises, with the equipment linked together in much the same way that distributed workstations are connected via local-area and wide-area networks.

These links potentially will need to carry massive amounts of data at speeds greater than 1 Gbit/sec across a campus environment. Sounds like the perfect application for fiber optics, doesn`t it? After all, you might think, they don`t call it Fibre Channel for nuthin`.

Well, not so fast, say sources. Copper cabling dominates the existing environment. And at least one source sees that dominance continuing, at least for the next few years. "The projections [for fiber] look good--but they`ve looked good for a while because everyone has always said that optics is going to replace copper," the source pointed out before delivering the capper: "But as long as the transceivers stay at the cost that they are..."

If this refrain sounds familiar, it should: It`s the same dirge-like soundtrack that has accompanied fiber`s efforts to reach the desktop and the home. It seems that every time fiber proponents attempt to extend the technology beyond its current domination of long-haul carrier backbones, they find themselves tripped up by the issue of electronics costs.

The cost-sensitivity of the Fibre Channel marketplace should only add to the heat already being felt by optoelectronics manufacturers. As might be expected, these suppliers are attempting to lower the price of their products--with varying degrees of success. The problem, said the vice president of one optoelectronics firm this past summer, is that the business of making fiber-optic components is too much like portrait painting or tailoring: It relies too much on the expensive and time-consuming hand work of skilled individuals. Until it outgrows this "cottage industry" ethos, he said, the optoelectronics community will never provide products in the quantity or at the price necessary to attack the home and office markets.

He could have added "network attached storage" to the list.

Help is on the way, however. I recently visited a West Coast firm that is working to automate the optoelectronics manufacturing process. There I saw systems that, through a combination of precision machining and monitoring equipment capable of providing a clear look at ridiculously small joints and connection points, promise to reduce to 15 min assembly procedures that now take hours by hand. This firm is not the only one in its field, and tools that would significantly expedite component manufacture should soon be available from a variety of suppliers.

Such advancement won`t come cheaply: The assembly system I saw retails for more than $250,000. But optoelectronics manufacturers will have to make such investments if they want their markets to grow. Although long-distance and high-speed requirements have always favored fiber, copper system manufacturers are constantly working to close the gap between requirements and performance. xdsl is but one recent example. As the slow growth of fiber-to-the-desk and fiber-to-the-home markets attests, many users appear willing to settle for "good enough for now" because future-proofing with fiber seems just too expensive.