When it comes to advancements in structured cabling and networking technologies, there is always something bigger, better and brighter just down the road. Here's a look at what may be around that next corner.
March 1, 2001
There may not be light at the end of the tunnel after all. If by “the tunnel” you mean the continuing quest for more and more network bandwidth to the desktop, it is beginning to look as though the light waves used to carry data over optical fiber may never replace the electrons used to carry it over copper wire.
“With every generation in copper technology, the theory is that well, the next step is going to be fiber,” says Robert Kostash, sales director for communications solutions at Avaya Inc.’s Canadian office in Toronto. Yet successive generations of technology have squeezed more and more out of twisted pair, and the fiber take-over that once seemed inevitable, has been put off. “I’m beginning to be of the opinion that it’s probably never going to get there,” Kostash says.
Some might disagree. The Tolly Group, a Manasquan, NJ research firm, published a white paper last July arguing that fiber could soon be the preferred medium for 10- and 100-megabit-per-second (Mbps) Ethernet connections to desktops.
The research firm based its view on cabling architecture. Because Category 5 copper cabling can only carry Ethernet traffic about 100 metres without amplification, a building network typically needs a wiring closet on each floor. These closets are then connected by a vertical backbone — quite often using fiber. However, the Tolly Group argued, fiber can go 300 to 500 metres at a stretch, so there is no need for the multiple wiring closets. Instead, fiber can run from the desktop to a patch panel on each floor and then carry on to a single wiring closet.
This eliminates the space needed for multiple wiring closets, and cuts down on equipment. Kostash notes that one advantage of this model is that it avoids wasting ports on multiple switches and routers. Many building networks have hardware sitting in every wiring closet with empty ports, whereas if all the gear is in one place, there should never be a need to have more than one unit with unused ports.
The Tolly Group study — commissioned by optical fiber manufacturer 3M — argued that if this “home-run” cabling approach is used, installing fiber can actually be as economical today as installing copper.
But Kostash, while he understands the reasoning, says his experience is otherwise. There are indeed savings to be had from designing the network around fiber’s characteristics, he says, but “our experience has been that even when you go through that argument and you do that design and you put in the cost of the network electronics, it’s still more expensive than a horizontal copper, fiber backbone design. While it’s getting closer, it’s still a higher cost.”
The analyst who wrote the Tolly Group report has since left the company, and despite several attempts no other spokesperson for the firm could be found to discuss the paper.
If economics can’t make fiber prevail over copper to the desktop, the other thing that might is bandwidth demands. Today, desktop connections are generally 10 or 100 Mbps. Gigabit Ethernet, which increases network speed by another order of magnitude, is gaining popularity in the backbone. So far, Gigabit Ethernet runs almost exclusively over fiber.
There are efforts to support the higher speed on twisted-pair, but distances are extremely limited and the technology is largely untested, says Rich Seifert, president of Network & Communications Consulting in Los Gatos, CA. There will probably be a market for Gigabit Ethernet to the desktop eventually, Seifert says, assuming it works — but nobody knows that yet because there is not enough volume to say. However, Seifert believes Gigabit Ethernet is more likely to come to the desktop via improved copper technology than over fiber.
In the meantime, copper cabling technology continues to improve, albeit slowly. For North American customers, the latest step forward is the Category 6 standard for twisted pair. Category 6 will provide greater bandwidth than the currently popular Category 5e — up to at least 200 megahertz (MHz), or 250 MHz with crosstalk cancellation capability installed, versus 100 MHz — in addition to improvements in attenuation, crosstalk and return loss. The net effect of all this is that Category 6 cable will handle more data, more reliably than Category 5e.
It is not happening yet, though. Kostash says that most new installations right now are still using Category 5e cabling. The main factor holding up adoption of Category 6 is that the standard has not yet been formally ratified. “Some people, particularly government, are reluctant to commit to a standard that’s still not published,” Kostash says. That should happen some time this year, and once it does, industry observers expect the new standard will quickly gain popularity in new installations.
Optical fiber is considered more secure than copper, because it is harder to tap, and that appeals to some government and military organizations and others with stringent security needs, such as defence contractors. And then there is the “future-proofing” angle — organizations wiring up new premises where they expect to stay for a long time may think it prudent to install fiber on the theory that they will eventually want it. Still, Kostash says he is only aware of one all-fiber network installation in Canada in the past two years.
Rob Stevenson, communications division manager at Guild Electric Ltd. in Toronto, agrees that “in most organizations there’s still an awful lot of headroom on the copper they’ve got.” He sees no immediate pressure for fiber to the desktop.
Then again, in technology the word “never” should be used very cautiously. If the cost of fiber and associated equipment keep dropping, it could still win in the long term. And there are a few unknowns that could be significant. Amy Copley, senior product manager at Sycamore Networks, Inc. in Chelmsford, MA — which sells its optical switching products to carriers rather than enterprise customers — points to this year’s electricity shortages in California and notes that one advantage of optical networking is lower power consumption. “It makes a lot of sense, if you’re optical all across the core, that the fiber will make a deeper and deeper reach,” says Copley.
Whether or not it will ever make it to the desktop, there is no denying fiber’s role in the enterprise network. Optical technology is well established in the backbone and that will only increase. And in that realm, Stevenson says, the area to watch now is the arrival of thinner optical fibers that can carry a signal farther without amplification.
Optical fiber is not like, say, water pipes in this respect. Making a pipe thinner would reduce the amount of water that could flow through it. Making an optical fiber thinner does not reduce its capacity, but does allow it to carry a signal farther. The reason is that a beam of light travelling through an optical fiber rarely travels in a straight line through the fiber. Instead, it constantly bounces off the outer cladding of the fiber. The thicker the fiber, the farther the beam of light must travel to get from one end to the other.
For some time, 62.5-micron fiber has been the most common, Stevenson says, and this can carry Gigabit Ethernet connections somewhat less than 300 metres. Now some fiber manufacturers are producing 50-micron fiber, which will support Gigabit Ethernet over runs of 300 and even 500 metres.
Just how significant this is remains to be seen. Most office runs fall within about 90 metres, Stevenson says, so there seems to be no immediate need for longer-distance fiber. But that could change. “Nobody’s going to know what the driving technology is until it hits,” Stevenson says, “and then everybody will be driven.”
In the meantime, the real network bottleneck is not within the enterprise itself but in the so-called “last mile” — the local loops connecting the organization to the central-office switches of the public carriers.
Ever-increasing Ethernet speeds have brought more capacity to LANs, while fiber has bulked up the public network to keep pace
with the demands created by growing Internet use and other needs. The problem is tying the two together. One possible solution is passive optical networking.
The old way to bring fiber from the central office right into an office or home was to run a single fiber from the central office to each customer. This was expensive — too expensive for homes and smaller businesses. Passive optical networking reduces the cost by running a single fiber to a group of customers. The fiber goes to a splitter, which divides signals among multiple fibers running into the customer premises.
The connection is still fiber all the way, but much less fiber is needed. It is like car-pooling: everyone goes a short distance to one location and then gets together to share one vehicle for the rest of the trip. The single fiber back to the central office can easily carry multiple signals.
“You’re slicing a wavelength to hit a smaller user, explains Todd Cabral, a spokesperson for Quantum Bridge Communications, Inc., an Andover, MA supplier of optical networking gear for the last mile.
Quantum Bridge supplies its technology to carriers, allowing them to offer access services to customers. Besides Internet access, Cabral notes, those services may include virtual LAN capabilities, permitting customers to tie together two or more of their own locations over an all-fiber connection that looks to everyone on the network like a single Ethernet connection. Another application is building integrated voice and data networks linking multiple locations, Quantum Bridge officials say.
CONSIDERING JET FIBER
Micro conduit, or jet fiber, is another development that could make a big difference in the last mile. It is expensive and time-consuming for carriers to run fiber to customer premises when the customer orders it, yet not economical to install fiber everywhere in hopes that customers will choose to use it. Micro conduit solves this problem. A four-inch plastic conduit can contain tens or hundreds of tiny stainless-steel tubes. Once this conduit is installed, the carrier can blow in fiber when a customer orders it, without disturbing the conduit. According to Bill St. Arnaud, senior director of network projects at the Canadian Network for the Advancement of Research, Industry and Education (CANARIE) in Ottawa, fiber can be blown into micro-conduit for distances as long as 20 kilometres — and thanks to the stainless-steel micro-conduit, raw fiber can be used.
Moving into the core network itself, improving optical technology promises quicker and easier access to long-haul bandwidth. Optical switches are just beginning to replace cross-connects that have to be reconfigured manually to set up circuits for customers. The optical switches can be programmed remotely, explains Copley at Sycamore Networks, which means a carrier can respond to a customer’s bandwidth needs within days instead of months, and some day soon might be able to provision a circuit almost instantly.
OPTICAL SWITCHING ON THE RISE
Communications Industry Researchers (CIR), a Charlottesville, VA research firm, expects U.S. carriers’ spending on optical switches to grow from US$234 million last year to some US$7.4 billion by 2004. CIR also forecasts that while today’s optical switches still generally use an electrical core — converting data from optical to electronic form for switching and then reconverting it into optical form to send out onto the network again — there will be an eventual move toward all-optical switches as growing network demands outstrip the optical-electrical-optical (OEO) switch design’s capabilities.
Multiprotocol lambda switching (MPLS) will provide more control over the paths that network traffic follows through the network. St. Arnaud describes it as “a way to turn packet networks into circuit-switched networks,” noting that traditional carriers like the concept, while reviews within the Internet community are mixed. By controlling the routing of network traffic, MPLS may help guarantee quality of service, which is particularly important for voice and video traffic.
In cabling and networking, predictions are risky. If fiber to the desktop has taken longer than expected to arrive, it is largely because copper cabling technology has improved far more than we once would have thought possible. Meanwhile, optical technology has also advanced, opening up possibilities that would have seemed unlikely until recently. And it is fortunate that both these things have happened, because applications that would have seemed outlandish in the early 1990s are now widespread, bringing with them demands on networks that were not imagined a few years ago.CS