My December column covered the development of copper cabling standards through the 1990s and the emergence of optical fiber as an alternative due to its higher bandwidth and distance capability. While one must admire the technical efforts of copper cable and transceiver manufacturers to develop higher speed links, the new technology created in the later 1990s for fiber optic links up to 10 gigabits per second were simply amazing.
It took two major efforts to get multimode fiber optics up to gigabit and faster speeds. First, it required development of a completely new type of optical source. Light-emitting diodes (LEDs) were used on most lower speed multimode links, but LEDs ran out of steam at about 250–300 megabits per second.
A new laser, the vertical cavity surface-emitting laser (VCSEL), provided plenty of bandwidth and power for faster links. It also had one more compelling feature—an extremely low cost. However, VCSELs could only be made with outputs at 850 nanometers, so multimode optical fiber had to be redesigned to provide higher bandwidth at that wavelength.
An old fiber design with 50/125-micron core/cladding was chosen to be used with VCSELs, since it worked better with lasers. Over the following few years, this new fiber, designated OM2, was developed into OM3 fiber with more than 10 times as much bandwidth capability as the old 62.5/125-micron (OM1) fiber, and eventually, it was developed into OM4 fiber with 20 times as much bandwidth.
Once again, fiber outshone copper cabling. At gigabit Ethernet speeds, these fibers could allow links five to eight times longer than unshielded twisted-pair (UTP) copper. When 10G Ethernet was created a few years later, fiber links were available for several years before copper and provided three times the link distance.
For gigabit networks and above, fiber provided another advantage—very low power consumption. The VCSEL-based links required only a fraction of the power required for copper links.
With all these advantages—longer links, lower power and early availability of commercial products—one would think that fiber would have finally won the battle over copper. But it didn’t happen. Part of the reason was users’ comfort with copper and skepticism over using fiber. “It’s glass. It’s so fragile, hard to install and expensive!” they said. The biggest problem was probably the cost of converting copper ports to fiber. Every connected device had a free UTP connection; connecting to fiber required a media converter costing about $100. Plus the big network switches charged $500 or more extra for a fiber port.
While this big fiber/copper battle was going on, with each side focused on the other, pointing out their weaknesses, something big happened right under their noses—wireless.
Part of the wireless takeover of the network was due to the mobility it allowed. In the beginning, users were willing to forgive the slowness of wireless because of the freedom it offered, but they kept their desktop computers on Cat 5 for when they needed more computing power and faster networks. Then laptops got more powerful, and Wi-Fi became a free feature, often replacing the old Cat 5 port. Finally, Wi-Fi got fast enough that it was often hard to tell if you were on a wired or wireless network.
Laptop computers soon surpassed desktops in sales and are now being challenged by tablets. Smartphones began using Wi-Fi as well as cellular connections, and network managers who had once looked upon their cabling as an essential asset of their company began to focus on getting Wi-Fi coverage to everyone. Cabling was often relegated to connecting power users such as engineering or graphics and, of course, connecting all the Wi-Fi access points the company needed.
That’s what’s often misunderstood about wireless—it’s not wireless. Wi-Fi only replaces the patchcord used to connect a device to the network. Every Wi-Fi access point requires cabling to connect to the network, and—a very important issue—it needs gigabit Ethernet to provide adequate bandwidth to users. The copper or fiber cable that once went to every desktop now goes to a Wi-Fi access point that serves dozens of users, meaning fewer cabling runs are needed.
Today, much of the focus of premises cabling has moved over to supporting wireless antennas, not just Wi-Fi but also indoor cellular antennas to support all the smartphones and mobile devices moving around inside the building. Cabling standards have been slow to respond to these changes; TIA-568 still hardly even mentions wireless.
Not only has wireless become the network connection of choice for the majority of users, but it also seems another competitor to copper and multimode fiber has begun capturing a big part of the network market. I will pick that up next month.