Taking fiber to the desktop may be only a question of when
Fiber vendors are confident they hold the answer to future horizontal network needs. Now if they could just convince a very skeptical customer base...
Robert Pease Associate Editor
Fueled by an ever-increasing need for faster speeds and greater capacity, fiber is making headway into applications once believed to be years away. Thus, as we anticipate another year of further research and development in the fiber-optic industry, network managers at every level face the dilemma of how far to go with fiber--to the curb, to the building, to the cabinet, or to the desk?
Of these four application areas, the last is generally agreed to be the most resistant to fiber penetration. The architects of local area networks (LANs), particularly in the corporate arena, want the most efficient, future-proof, and economical system possible. Thus, one of the biggest questions they face is how much of a cost premium, if any, they need to spend on fiber to ensure a long-lasting, scaleable, integrated network.
The fiber industry has seen its technology reach into wide-area and access networks and increasingly become the medium of choice for campus and enterprise backbone applications. In each instance, fiber has crossed the boundaries of applications once dominated by copper and hybrid architectures. Not surprisingly then, there has been little argument among the vendor community over whether corporate fiber managers will see the wisdom of investing in fiber to link the backplanes of their desktop computers. The biggest debate seems to center on when that decision will become commonplace.
Fiber or copper?
It isn`t difficult to find members of the fiber-optics camp who believe copper technologies have already begun to lose their domination in horizontal-cabling applications to fiber-to-the-desk. "Just look at the number of manufacturers once considered `strictly copper` manufacturers that over the last two years have entered into the fiber business," says Tony Beam, director of global systems marketing at AMP Inc. (Harrisburg, PA), one of a handful of companies that already market fiber-to-the-desk systems. "Whether they`re into cable, connectors, or connecting hardware, it does verify the fact that they feel threatened to some extent. The industry of LAN electronics has always been the hold-out, except for some second- and third-tier suppliers, but recently fiber-to-the-desk seems to be penetrating that industry."
At Siemens Microelectronics Inc. in Cupertino, CA, where fiber-to-the-desk is already being marketed as the replacement for Category 5 copper, the belief is that with all the advantages fiber offers over copper, it isn`t just a threat in replacement applications...it`s a reality. Citing the advantages of practically unlimited bandwidth, no noise interference problems, smaller size, more flexibility, elimination of ground loops, longer distances, and lower cost than Category 6 or 7, sources at Siemens insist fiber is the way to go when replacement opportunities present themselves.
"If you`re thinking of replacing your network, fiber-to-the-desk is the way to go," asserts Jerry Sheridan, a fiber-optics business-development manager at Siemens. "Anytime the complete transformation of the network infrastructure is taking place, a long-term approach of supplying enough capacity for the next 15 years deserves consideration. Fiber-to-the-desk allows a maximum of benefits when the option is chosen to integrate the telecom and datacom services in the same network."
Dan Silver, product market development manager at 3M, adds that the fiber-to-the-desk advocates are successfully sending this message to new builders as well. "We`re telling them they can be ready today and don`t have to wait for that killer application to put large bandwidth demands on their system," says Silver. That`s because the killer app is just around the corner. "Gigabit Ethernet is going to promote fiber-to-the-desk more than anything else that`s happened in the last two to three years," he predicts.
The Gigabit Ethernet specification standard, written and released last July, is only for fiber at this time, points out Silver. Copper, he says, is lagging and that gives a "shot in the arm" to fiber`s appeal, especially when distances of more than 100 m are being contemplated (see Table 1).
Silver and other fiber-to-the-desk proponents are also working to lower the cost hurdles associated with fiber--and to ensure that users realize progress is being made. "People perceive fiber to be 400% more costly than copper in new builds," says Silver. "We in the industry haven`t done a good job of talking about the cost-effective fiber-to-the-desk systems that are out there today. But we`re working hard to get that message out, though it`s a tough perception to eliminate."
Silver and others don`t advocate pulling out in-place copper architectures if they are doing the job. On the other hand, if a retrofit opportunity exists, such as when gutting a floor or constructing a new building, he believes it`s an excellent time for users to reexamine the economics of fiber versus copper. "People need to be forward-thinking and realize those killer applications will eventually demand the higher bandwidth that fiber provides," says Silver.
Some applications already exist that require a serious look at fiber-to-the-desk technologies, say advocates. For example, videoconferencing, computer-aided design, graphics, pre-press, and other image-intensive applications are driving fiber closer to the desk in local networks used in hospitals, college campuses, and even some corporate environments. Although the fiber may not be physically connected to personal computers in all cases, it`s reaching steadily in that direction.
DWDM and fiber-to-the-desk
Probably the most significant development in recent years for the telecommunications industry is dense wavelength-division multiplexing (DWDM), the process of using multiple wavelengths on a single fiber strand to dramatically increase its capacity. DWDM was immediately embraced by long-haul telecommunications networks and, during the last few years, has rapidly advanced toward the premise and corporate campus network environments.
As DWDM applications move into smaller corporate environments, they will push fiber-to-the-desk technology, says Dana Hardgrave, director of marketing and business development for optical networks at Alcatel. "Delivering end-to-end wavelength services is what the optical layer is all about," says Hardgrave. "With DWDM, more channels can be carried over one fiber, and as metro DWDM solutions begin to prove in, more wavelengths will be available in the network for providing such services as fiber-to-the-desk."
Lucent Technologies (Murray Hill, NJ) placed a metropolitan DWDM system in Microsoft`s campus network in Redmond, WA, and believes this deployment is a definite sign that using multiple wavelengths in fiber-to-the-desk technology is not far off.
"With carriers starting to look at DWDM as a means of boosting capacity and supporting more data traffic on the local access side, optical networking will give them a means of serving individual corporate customers," said Kathy Szelag, vice president of Lucent`s Optical Networking Group. "With DWDM, these carriers will be able to lease bandwidth, via wavelengths, to businesses that will in turn be able to deliver this capacity to every office. The result will be fewer network bottlenecks at the desktop computer level. Already, we`re seeing companies like Microsoft experiment with sending a wavelength to each department."
Vishwa Narayan, an applications engineer for Ericsson`s components division, agrees that DWDM, along with optical networking technologies, will enable terabit transmission networks and make it feasible to have gigabit access technologies.
However, 3M`s Silver disagrees, at least for now. "I don`t hear or see anything about DWDM on a campus or inside of a building," says Silver. "We don`t, at this time, see a need for DWDM in fiber-to-the-desk technology. I may be the one being short-sighted, but I don`t see a need for it now."
Others agree that DWDM, although certainly a driver for larger fiber applications, is not likely to influence decisions for fiber-to-the-desk applications. Stephen Barr, business manager for Nortel (Brampton, ON, Canada) points out the high cost of the electronics associated with DWDM. "The lowest cost and simplest situation will always win," says Barr.
At Siemens, Sheridan also believes DWDM is still a long way off from being needed for horizontal builds to the desktop. "Fiber alone has more than enough bandwidth to support desktop applications for the next decade," says Sheridan. "Any additional passive components will only add complexity and cost."
Not so fast
Complexity and cost have been traditional arguments against fiber in horizontal applications. And for every proponent of fiber-to-the-desk who says the future is now because these barriers have come down, another industry voice will contend that fiber-to-the-desk`s future is still just that--in the future.
At Alcatel USA (Dallas, TX), senior product line manager Robert M. Keating believes that today`s typical copper infrastructures meet the existing service needs of commercial and residence customers just fine. "Our estimate is that it will be at least five years before the equipment cost [for fiber] will equal the value provided to the customer. Only when some service that the consumer desires can only be provided over fiber, or when the initial first cost of the equipment is on parity with or cheaper than copper, will fiber be the way to go," says Keating.
Jim Hayes, president of Fotec Inc. and director of Fiber U, feels it is doubtful that bandwidth requirements will overwhelm most copper infrastructures in the near future. "Fast Ethernet or FDDI [Fiber Distributed Data Interface] at 100 Mbits/sec is adequate for practically every need." However, "Gigabit Ethernet will be a backbone technology, almost exclusively, and it will be predominately fiber," he concedes.
Hayes also sees a major breakthrough on the horizon for fiber-to-the-desk advocates. The new short-wavelength 10/100 Ethernet standard using 850-nm light-emitting diodes provides a migration path from 10 to 100 that was previously very expensive (see "Push for short-wavelength Fast Ethernet standard," Lightwave, July 1998, page 17). This more logical standard for desktop applications using fiber, says Hayes, provides parity in the cost for cable plants of fiber or copper. "It`s the electronics that still make copper more palatable," he says.
Similar thinking leads Ericsson`s Narayan to feel confident that fiber-to-the-desk technology is still at least three years away in the corporate environment. He believes corporate demand for bandwidth will be a driver for fiber-to-the-desk, but cost remains a major deterrent.
"Corporations with deeper pockets and the need for business-driven bandwidth are probably ready for fiber-to-the-desk today," says Narayan. However, he adds, "Today, copper telephone lines imply bandwidth in kilobits. xDSL [digital subscriber line] promises megabits. Coax [for hybrid fiber/coaxial-cable applications) promises mega bits. Fiber promises gigabits. But I think we need to go from kilobits to megabits before talking gigabits." That`s particularly true for horizontal applications outside the corporate environment. "What home applications do we have today that demand gigabits?" he asks.
And even if technology advances dramatically in the fiber-to-the-desk area, how much will the average consumer be willing to pay for it? Narayan predicts many years will pass before the first home applications demanding gigabit capacities will appear. But he concedes that eventually, in a decade or two, fiber will be the natural winner in all applications.
Future for fiber-to-the-desk
If consensus among industry sources on the timing of fiber-to-the-desk is difficult to achieve, the analyst community appears more unified--but not completely.
"We may be within a year or two from seeing measurable fiber-to-the-desktop for high-performance applications," says Larry Vanston, president of Technology Futures Inc. (Austin, TX). "Several more years will go by before it reaches the mainstream. Then, we would likely see a phase of fiber replacing copper on a large scale as most users go to Gigabit Ethernet or Asynchronous Transfer Mode."
Vanston does predict some holdout applications that may take longer for fiber to penetrate. In the home, for example, he believes speeds of less than 100 Mbits/sec will be adequate for at least another decade.
"To me, it comes down to bandwidth requirements," says Vanston. "If you have them, you`ll look to fiber. Otherwise, Category 5 cable looks pretty good."
As mentioned earlier, another con sideration for fiber versus copper is whether the buyer is faced with a "greenfield," or new build, scenario or whether cabling infrastructure is being replaced. Robert Rosenberg, president of Insight Research Corp., a Parsippany, NJ, telecommunications market-research company, has observed the battle lines being drawn. Copper`s "ace in the hole" is still the fact that as long as existing infrastructure meets the end-user criteria for speed and bandwidth at the desktop, it`s unlikely to be replaced in the near term. However, in greenfield situations, giving consideration to fiber-to-the-desk is more advisable.
"If I were building a new building, knowing what I know about the trends in telecom, there`s no question at all that I would stuff the raceways with singlemode fiber and run it all the way out to the desk," says Rosenberg. "Even if it`s perceived as overkill right now, my building would be assured of a very, very long life...about 40 to 50 years. It`s not just a five-year play."
Rosenberg sees today`s fiber-to-the-desk market as relatively small in comparison to other telecommunications markets--under $1 billion as far into the future as 2001 (see Table 2). But he`s comfortable saying the enterprise and corporate environments are headed for fiber-to-the-desk technology within three to five years.
There are several application and price dynamics that will speed optical fiber`s reach to the desktop, says Stephen Montgomery, president of ElectoniCast Corp. (San Mateo, CA). Small- form connector/transceiver packages will drive down the cost of LAN applications, he says. Vertical-cavity surface-emitting laser advancements, along with their lower prices, will enable the re-emergence of 50-micron multimode optical fiber to accommodate Gigabit Ethernet. Options at 62.5 microns are available for Gigabit Ethernet as well. Other technologies, such as media converters that enable optical fiber to be used with existing electronic hubs or switches, airblown-fiber systems for growing indoor fiber networks, and fiber`s immunity to interference problems, will help make fiber an attractive alternative for future-proofing horizontal cabling networks, he predicts.
"The major cost in any network is the actual installation of the cable," says Montgomery. "So if future applications for networking demand more bandwidth, the LAN administrator will be replacing Category 3 with Category 5, then Category 5 with Category 5E, and on it goes. The alternative is to install, just once, optical fiber that can accommodate high-bandwidth demands."
Eric Lach, a network technologies research analyst with The Gartner Group (Stamford, CT), agrees that horizontal cabling will eventually become a pro-fiber environment, but believes it will likely take at least a decade to get there. Lach says Category 5 is the most widely accepted, well understood, best documented, and most standardized medium for horizontal cabling in most enterprises today. The "comfort factor" for copper is well established, despite a hardcore group of cabling vendors who are working diligently to make fiber more palatable to a wider range of corporate customers.
"This group has worked on each of the objections that people have to using fiber, such as installation difficulty, fragility, time consumption, no standard connectors, expense, and a lack of network cards," says Lach. "They`ve directly addressed each issue and have come up with some good solutions for the most part. But while pushing to advocate fiber for prime time, they`ve also highlighted their dissent and lack of consensus within their industry. Despite that, however, they`re moving forward and deserve the credit for their progress."
Lach believes the battle between copper and fiber will reach two "forks in the road." The first, in about five years, should see fiber cards available and costs reasonably comparable. At that time, new builds will likely find fiber more attractive as copper begins to run out of bandwidth steam. At that time, Lach believes fiber will gain some ground to the desktop. It`s at the second fork, in about 10 years, that more people will look down the road and see the need for more bandwidth coming. As they look ahead another 10 years, fiber will likely get the nod. But it isn`t a certainty that copper will yield its turf easily.
"The copper people have pushed copper well beyond what most people ever believed copper capable of," says Lach. "They may even be able to add more life to it. But I still think we`re all eventually going to fiber. The only question is, when?"
Which leads us back to our original question. How far will corporate network designers be willing to go with fiber, now and in the near future? It`s probably fair to say that most people in the fiber industry say today`s best chance of getting fiber close to the desktop is in new-construction situations. As soon as enough network interface cards are available for fiber, these installations would include fiber all the way to the backplane of the computer for enterprise applications. For existing infrastructure, however, fiber may still be at least a decade away from winning support to the desktop, with the only exceptions being a handful of bandwidth-extensive applications.
As long as copper continues to provide adequate speeds and capacities for most business applications, network managers will have very little incentive to yank copper out of raceways in favor of new fiber. But if speed and bandwidth requirements increase as expected and more fiber-supportive technologies find their way into the horizontal builds, fiber will likely enjoy the majority of support as tomorrow`s transmit medium of choice...all the way to the desk. u