Fiber to the desktop is steadily gearing up for proliferation into the marketplace. But don't expect big things from it this year.
April 1, 2002
Optical cable has become de facto in long-haul communications networks, as predicted. Local service carriers are deploying it extensively in metropolitan area networks (MANs). Optical cable is also commonplace for short-throw links between high-speed network and storage devices, and for patching high-speed line cards or linking communications chassis.
But what about optical in the local area network (LAN)? Will we finally see fiber to the desktop (FTTD) in organizations everywhere — replacing the now familiar twisted pair?
The answer, again this year, appears to be yes, but not now.
For those on the copper side of life, fiber is a whole new ballgame. Optical cables look familiar, but contain individually insulated strands of glass rather than copper. Data travels on LASER light, not riding the familiar electromagnetic and radio frequency (RF) signals carried across more common copper links.
And herein lies strike one against fiber in the common LAN. Optical is not electromagnetic, so everything changes: different rules, different gear, different installation specs, different test and measurement systems, and different management and administration techniques. Even network design becomes a whole new discipline.
Strike two is cost, which most industry pundits place at levels several magnitudes above copper-based solutions.
Strike three, so far, appears to be need.
So, running fiber to each and every desktop is not a minor upgrade. But fiber has several attractive attributes as well. Optical handles distance better than copper, carrying data much further and with far less frequent re-amplification of the signal. Twisted pair (in the LAN) “max.’s out” at about 90 metres from the cable closet (less than 300 feet), so the network electronics in a copper-based LAN must remain close to the desk. Fiber runs, on the other hand, require only passive patch panels locally, and can service single links up to 20 times that of copper.
But fiber has one other perceived advantage — security: no electromagnetic radiation (EMR) or RFs to tap. Unlike copper, you can’t just “clamp on” to a fiber — you need an experienced technician and expensive equipment. Needless to say, FTTD has its uses.
IS FTTD NECESSARY?
Organizations that rely upon high-resolution imaging applications often have call for the multi-gigabit speeds optical LANs bring to individual workstations, says Roberta Fox, president of Fox Group Consulting of Markham, ON. Land registry and museum applications, multiple parties consulting over detailed 3D medical images, and high-level computer animation are the examples she uses. But in most normal situations, says Fox, the potential speeds just are not there to justify the cost of installation and maintenance.
Twisted pair, on the other hand, is cheap, she says. And it is easy to handle. To top it off, Gigabit Ethernet, the mainstay FTTD solution for those who do require high bandwidth, is now more easily available over copper, using standard structured cabling.
“We’ve had clients ask us to look at fiber to the desk,” says Fox. “Particularly in government and healthcare — the two industries that still think optical may be the requirement. But it takes more skill to install and maintain fiber. It’s more fragile. You can go gigabit Ethernet at far lower cost.” And, she points out, “you can walk on that.”
Fox says her neighbour, an engineering design firm with very heavy CAD (computer aided design) and graphics traffic over their network is a great example. The company just installed 100 MB Ethernet over copper — today’s most common, of-the-shelf, networking solution, she reports. “Their main issue is faster processors and lots of memory,” says Fox. “Not getting information to each other.”
Still, optical has its applications. And a slow ramp into optical LANs appears to have finally begun.
STUDY PREDICTS GROWTH
“Our research shows that fiber-to-the-desktop is no longer restricted to early adopters,” says Kathryn Korostoff, president of Sage Research, Inc. of Natick, MA, citing the firm’s recently published survey of US businesses. “It is moving into the mainstream,” she adds.
Yet a more sobering view develops. “Despite this growth,” says Korostoff in a brief analysis of the research results, “there are significant factors that may affect even wider adoption of fiber-to-the-desktop. These include improvements to copper-based solutions, the perceived relative high-cost of fiber optic cabling, and wireless alternatives.”
The Sage study also notes:
despite strong growth in multimode fiber use, Category 5 systems will remain dominant in the horizontal networks through mid 2003;
a strong preference for copper cabling, with many companies investing in next-generation copper products rather than choosing fiber. (A trend that Sage notes is evident across all seven years of studies.); and
a “strong interest” in Category 6 products among those organizations currently planning cabling upgrades.
Despite these hurdles, Sage predicts a rosy future for FTTD, and suggests a rapid growth will begin soon. The company notes that nearly 25 per cent of organizations currently have multimode fiber in their horizontal networks, and predicts that number will rise to 33 per cent over the next two years.
Take note, though: Sage’s 1995 report delivered very similar predictions that proved highly optimistic.
What is new this year, however, is a sense that change is closer. Still, don’t be trading in your wire cutters for splicing blocks just yet.
Nick Tidd, Managing Director at 3Com Canada in Mississauga, ON concurs: “This is an interesting debate in that corporations recognize that at some point in the future fiber to the desk will become a necessity. But it’s based upon provisioning bandwidth to the desktop to optimize applications,” he points out, “and the point people continue to drive home is that they just don’t need fiber today.”
It is on the RADAR screen, says Tidd, but it is just not being implemented until necessary. “Most corporations are looking today at what they can do with their existing infrastructure,” he says. “So they would rather stay with copper at this point.”
And it will be a while before business sees a case for it, says Fox. “It’s about $125 – $150 per move, per MAC, for copper, but about $275 per fiber move,” she says. “A time domain reflectometer (TDR — a common piece of test gear for fiber) costs $50,000 plus,” she adds. “We’re not talking $400 Ethernet testers here! Just certifying your plant costs a lot more money. And the connectors are ten times the price.”
To top it off, says Fox, in some areas you may even have to deal with a different trade union to install the fiber.
“Fiber to the desk means fiber all the way through. But does everyone in your business need it?” asks Fox. “Even in engineering CAD design firms, worst case is only 60 to 65 per cent of the people requiring that kind of bandwidth to the desk. Are you going to install two infrastructures? One for the business people and one for the engineers?”
And it is not just the fiber itself, she adds, listing several different concerns that develop with a change to optical: network segment design, switched versus routed, how many people are on each segment, building design, and so on. “Everything changes,” she adds. “And almost everything costs lots more.”
On the defence, common structured cabling standards such a TIA/EIA-568-A and -B may be applied today to both copper and fiber installations in the LAN, making migration from copper to fiber seem somewhat less painful. But many experts maintain that while this is very handy, 568 was designed for copper-based networks. They say new schemes designed ground-up for FTTD installations bring far better economies to optical LAN deployment. But again, two to three years becomes the common theme when asked when the sweet spot will emerge.
Back at the ranch, even optical’s inherent security advantages are being questioned.
“I think there is a perception that fiber optics is more secure,” says Fox. “B
ut if there’s going to be in-building intrusion — if people really want access to the information — then they will have the TDRs, the training technicians and the ability to intercept. There may be a brief hiccup or loss of service while somebody puts a TDR sniffer on the line,” she adds, “but that’s not difficult to cover. And don’t forget,” she points out, “there is technology today that allows sniffers to pick up emanations directly from a computer screen and decipher them,” so the fiber/copper debate becomes a moot point in those applications requiring the tightest security.
“If sophisticated folks want to get into a place, a fiber optic infrastructure will not stop them,” Fox emphasizes.
Still, interest in FTTD is picking up, if only from marketing hype aimed at replacing business lost to recent global market developments — the meltdown of several high-profile optical carriers among them.
SLOW AND STEADY
A sure, but slow build appears to be in the wind.
“Server farms and aggregation are all going fiber,” says 3Com’s Tidd. “We’re seeing that today. And business is certainly evaluating it,” he adds. “But they’re putting it into their project plans for the next 24 to 36 months, because copper will carry most business plans over the next two to three years and they simply don’t see the need today. So until cost comes down and applications begin needing that level of provisioning, we are going to continue to see current technology stay in place.”
What happens then?
“Within two to three years you’ll begin to see applications and convergence happen at the desktop,” explains Tidd. “We’re already seeing the migration of voice systems onto IP. And 37 percent of all enterprise voice systems sold Q1 this year, were IP based,” he adds, citing a recent Phillips Infotech study (www.phillips-infotech.com).
“So the bandwidth needs are climbing rapidly. When you start to move voice, data and video traffic to the desktop, the requirement for bandwidth is going to become key,” adds Tidd. “And as we begin to see those transitions happen, the cost point will begin to drop. It may not necessarily become a requirement for everybody, but it will become a requirement for most.”
But it is Roberta Fox who cuts to the chase: “I believe fiber to the desk has limited application, but that those people that want the capacity and want the security will pay the higher up-front costs and the higher adds, moves and changes in exchange for what they believe to be important,” Fox says. “There is no doubt fiber is still suitable for the backbone,” she adds. “But to the desktop? Not yet. Not in the mainstream.”
Chris Blythe is a broadly published writer and infocom analyst. Send correspondence to firstname.lastname@example.org.
GUIDE TO FIBER TYPES
MULTIMODE and SINGLEMODE FIBER are the two types of fiber in common use.
Multimode fiber has a bigger core (almost always 62.5 microns — a micron is one one-millionth of a meter) and is used with LED sources at wavelengths of 850 and 1300 nm (for LANs).
Singlemode fiber has a much smaller core, only about 9 microns, and is used for telephony and CATV with laser sources at 1300 and 1550 nm.
Plastic optical fiber (POF) has a large core (about 1 mm) and uses visible light at 650 nm.
(Courtesy: Cable U – Lennie Lightwave’s Guide to Fiber Optics – 1996-2000, fotec, inc., Medford, MA – www.fotec.com/lennie.)