Will optical fibers to the desk ever see the light of day?

To many data communications people, no goal is more important (and seemingly more elusive) than Fiber To The Desk (FTTD). FTTD is the ultimate in performance. Once this goal is widely achieved, new high-bandwidth services will be available. Video-on-demand (the ability to watch any TV show, movie, or sporting event you want, at any time you want; along with the ability to edit, replay, slowdown, speedup, and remove commercials) would certainly be preeminent among these services. It would also forever alter the television and entertainment industries.

Though FTTD, along with high-bandwidth services, has been slow in coming, time may be right for it once again. Here are some of the reasons why this time may be different.

Copper data wiring gets ugly. The new varieties of copper data cabling (Cats. 6 and 7) are problematic to say the least. To put it briefly and bluntly: The standards that exist are not real standards. Cable and device manufacturers are using different bandwidth (signal speed) figures, the testing equipment costs a ton, and the tests get more complicated all the time. And if you ever thought there was any built-in "slop" in Cat. 5 installation requirements, it's gone now. To make matters worse, the new Cat. 6 wiring will be 20% to 60% more expensive than Cat. 5 wiring (depending upon whose figures you believe).

Fiber is cheaper. Don't believe me? Let me show you why. Those of you familiar with the EIA/TIA 568 standard will recognize a typical layout. Note how many hubs there are in this design; one on each floor of the building. These hubs are necessary for one reason: Copper wiring has severe distance limitations. That is, if you try to run copper data links for more than 100 m (including vertical portions of the run and all patch cords), the network will not work.

Now, consider a typical fiber-optic installation. This network uses fiber-optic cables to create what's called a collapsed backbone architecture. If you could see a drawing of it, you'd see only one hub. This is a big advantage with a fiber network (fiber has little difficulty with longer distances). And don't miss the ancillary savings this architecture provides: No power circuits are required for the hubs, equipment racks, backboards, UPS systems, conduits, and other small items. In addition, you also eliminate the labor for these items.

If hubs aren't required on each floor, neither are the Datacom closets. You can eliminate these dedicated spaces. This cuts down on cost and frees up space for other uses.

Does reality matter? And if so, when? Please don't pass this question off as frivolous; reality often takes a back seat to "Everybody's doing it," "I've never seen that before," "Don't make waves," and a hundred similar slogans. If reality mattered all that much, people would have switched over to fiber the first or second time they had to replace a three-year old copper data cabling system.

We've now gone through proprietary cabling systems, Cats. 2, 3, 4, and 5. Now, we are moving on to Cat. 6. Yet most people think fiber is too expensive because it has a 10% to 15% higher initial cost. (You're not expected to think about the fact that it lasts for decades.) The initial premium for fiber-electronic components over copper electronic components is $100 to $200 per node. But in the final accounting, using the collapsed backbone architecture saves somewhere between $200 and $500 per network computer. The question facing us is not whether people should change over to fiber, but whether the cost of ignoring fiber has become too great to bear. It may be that it has, or will be very soon.

A new viewpoint. The old network paradigm (pattern) was a smart network center and dumb equipment at the ends. This is how the telephone companies were (and are) set up; with complex switches and call accounting equipment at the phone company facilities, and low-tech telephones on your night stands.

The new paradigm is quite opposite, with relatively dumb network centers and brilliant personal computers at the end of the line. Internet messages from one computer to another contain their source and destination addresses, and routing is done in hundreds of semi-smart locations rather than at the center. This arrangement is perfect for fiber because fiber's endless bandwidth makes up for network intelligence. Note also that this dumb network is a whole lot cheaper to use than the smart network.

The bottom line here is because the overall economics of the new data world dictate that more bandwidth is so valuable it's virtually inevitable everywhere. That includes up to and inside buildings. Many new data services will be coming along in the next few years. In the data-networking world, bandwidth is the mother of invention. Only fiber can give such bandwidth.

The final effect of this paradigm shift is fiber is moving into the office building one way or another. And as fiber comes into the building, networking people begin to get comfortable with it. Eventually, they will get comfortable enough to go all-fiber, which will eliminate the need for a fiber/copper interface in the main telecom closet. Although this is not a major savings, it's not entirely insignificant.

Is fiber to the desk inevitable? No, fiber to the end of the line is not inevitable, although it's probably as good a bet as you're going to get. The one technology that could displace fiber inside buildings is wireless data transmission. If true high-bandwidth wireless technologies can be developed, and if they operate at very low costs, they could take fiber's place. However, this is quite unlikely.

We've tried indoor wireless networks, and we've come up short for most uses. There are problems with speed, radio-frequency interference, and antenna placement. Some genius may come up with new ideas that can work together to make wireless networks a reality, but don't count on it.

In conclusion. The entire networking industry has so much copper inertia, it has become the accepted wisdom. Very few people are bold enough to question accepted wisdom until some great outside force overcomes it. In such cases, the conventional wisdom collapses quickly, and is replaced by new conventional wisdom. When will this happen? When the pain of using copper and the pain of not having bandwidth becomes great enough; and not before.




Sidebar: A Bit of Evidence on Fiber's Behalf

The need for bandwidth is pushing some companies to look beyond copper and implement a fiber-optic solution.

A company on the West Coast installed a city-wide network using unused fibers it leased from a local telephone company. Bypassing the expensive "last mile" between the last telephone facility and its offices (using the leased fibers instead), the company saved (by their calculations) $300 per mile of fiber per month.

This arrangement allows the company to avoid a number of "techno-hassles." Their comments, from a networking magazine, were: "We don't have to worry about sophisticated switches or service restoration equipment. There will be enough bandwidth for any service, and we get this for a fraction of the cost."

Using fiber means these people don't have to switch protocols (to ATM or SONET), but can transmit over the fiber in their existing protocols. This saves a lot of pain for network engineers.