Roland Schlehuber
Bell Atlantic Federal Systems
Springfield, VA
I was puzzled by the Point of View article by Roger H.D. Lacey from 3M Telecom Systems Div. (see "Fiber-to-the-desk will change the installation business forever," February 1997, page 50). It seemed overly vague. What technological breakthroughs did Mr. Lacey have in mind?
The article perpetuates the myth that the primary cost difference between fiber and copper is the fiber cable, connectors, and labor to install and terminate it. This is misleading. Advances in fiber termination and reductions in fiber cable prices have reduced the fiber-to-copper ratio from 3:1 to 3:2.
The primary cost differential is the network interface hardware (e.g., adapter cards, hubs, transceivers), where a 2:1 ratio exists between equivalent technologies--such as 10Base-T and 10Base- FL--and a ratio of 3:1 or higher exists for fiber- specific technologies. The main source of this difference is the components, which receive and convert light to electronic signals.
Most desktop applications operate well in a 10Base-T or Token Ring environment. By using switching technology or upgrading to 100Base-T, it will be possible to meet most future customer requirements.
Unless desktop software requirements exceed the bandwidth limitations of copper (100 to 155 megabits per second now; 622 Mbits/sec in the near future) or the cost of the optics themselves is reduced, I do not envision a major conversion to fiber-to-the-workstation installations.
Roger H.D. Lacey responds: Mr. Schlehuber is correct in saying that fiber cable and connector technologies are increasingly cost-competitive with copper, especially with emerging standards for fiber connectors. Similar advances in electronic integration for fiber, and reduced packaging costs for optical electronics, will also lower the cost differential in network-interface technologies. And on a full-life-cycle cost basis, cost differentials are further reduced due to fiber`s advantages in testing, transmission, and maintenance.
The larger issue is whether users will require, or demand, the bandwidth offered by fiber-to-the-workstation. Already it is being used by early innovators as the solution to their existing high-bandwidth needs, which are unmet by existing copper technologies.
At the same time--driven by more-powerful applications, richer and more-complex content, and dramatic growth in the number of network users--the demands for bandwidth are increasing exponentially. In fact, emerging Fibre Channel, Gigabit Ethernet, and atm [Asynchronous Transfer Mode] standards may preclude the use of copper. The data throughput of ffol [fddi Follow on lan] in a shared-media architecture is 2.4 Gbits/sec, while the bit rate for Fibre Channel over a switched-network architecture is 1 Gbit/sec. Such protocols, along with a dramatic growth in the number of network users, high-resolution graphics, and distributed workgroups, are already pushing existing limits and could very shortly saturate available bandwidth.
Also, the combination and convergence of all these applications will vastly accelerate demand for bandwidth. Just like those PC users who 10 years ago couldn`t imagine filling a 20-megabit hard drive, people will find limitless opportunities that will change our world and make full use of vastly expanded capabilities.