Fiber optic advocates used to point to the copper types and snicker. They’d note that, while copper cabling went through eight or nine generations of coaxial and unshielded twisted-pair (UTP) cables, one fiber would have sufficed. But then along came gigabit Ethernet, then 10-, 40- and 100-gigabit networks. Fiber manufacturers first revived a 1970s design, 50/125 fiber, with more bandwidth when used with the laser sources in gigabit networks than the old standby 62.5/125 fiber. Next, laser-optimized 50/125 fiber was introduced, creating a horsepower race among fiber manufacturers. This led to new standards (and nomenclature) for 50/125 fiber, OM2/OM3/OM4, along with enhanced versions of each variety that promised even better performance than the standard product.

There also is “bend-­insensitive” fiber that promises you can kink it and staple it without damaging the fibers or significantly affecting its loss.

It starts sounding like Category 5/5e/6/6a/7, doesn’t it? Now, the fiber user and contractor have to choose among more than a dozen varieties of multimode fibers for premises cabling.
Adding to the complexity of choosing a fiber today are the controversies being aired in the market. Can you really have longer cable runs than allowed by the Ethernet standard if you purchase higher performance fiber? Are these fibers compatible with legacy fibers? Are they compatible with each other, especially different bend-insensitive fiber varieties?

We’re not finished yet. As the standards for 40- and 100-Gbps Ethernet (GbE) developed, network speeds surpassed the capability of fiber optic data links. (Copper cabling was limited to short coaxial runs, but you know that will change someday!) Two solutions emerged, both using multiple 10 Gbps channels. On multimode fiber, a parallel fiber solution was adopted. This requires eight fibers for 40 GbE and 20 fibers for 100 GbE. Forty GbE needs one 12-fiber MTP/MPO multifiber connector per port, while 100 GbE requires two 12-fiber MTP/MPOs or a new 24-fiber version of the connector. Single-mode fiber, however, has a wavelength division multiplexing (WDM) solution, putting four or 10 channels on one fiber at different wavelengths that requires only one pair of fibers for each link.

Google, the largest operator of data centers in the world, has stepped into this controversy. It plans to jump from 10 GbE directly to 100 GbE, skipping 40 GbE altogether. Google, however, is planning to build servers and switches that have 20 ports per blade (a mechanical form factor used for servers). If the company uses multimode fiber, it would have to manage 480 fibers per blade, but single-mode requires only 40 fibers. According to a Google speaker at a fiber optic marketing meeting, the company decided to use single-mode fibers in its data centers.

So now we have even more confusion: single-mode or multimode? “You’re Terminated” (ELECTRICAL CONTRACTOR, February 2010) highlights passive-optical LANs (POLs) based on fiber to the home (FTTH) technology. POLs leverage the low-cost components used in millions of FTTH installations to cut the costs of LANs, even making fiber cheaper than copper for medium and large networks.

Fiber has always been popular in the backbone of local area networks (LANs) but has had a hard time being accepted in the connection to the desktop. Fiber is preferred for the backbone because of its higher bandwidth and distance capability, but most desktop connections really don’t need more than 100 Mbps, easily handled by Cat 5 UTP cabling or wireless. Probably the biggest deterrent to using fiber to the desktop is that all computers already have a network port connection for Cat 5, and more users are opting for laptops, tablets or other mobile devices with wireless, rather than wired, network connections. POLs solve that problem by connecting local switches with single-mode fiber but providing copper ports for connecting devices like PCs or wireless access points.

Things used to change slowly when it came to fiber optics, but no more. These issues are being debated in some testy standards meetings and publicly in the trade press, white papers and webinars. And you should never count copper out. The advocates for Cat 7 (Class F internationally) are pushing harder for a U.S. standard covering their shielded twisted-pair solution for higher speeds on copper.

These issues affect major decisions by end-users and contractors. As someone in the middle of the technical discussions and an industry insider, I must say that the issues are not black or white. Making decisions today is tough. Over the next few months, we’ll look at the issues with multimode/single-mode and bend-­insensitive fibers, and we may take another look at fiber versus copper.


HAYES is a VDV writer and educator and the president of The Fiber Optic Association. Find him at www.jimhayes.com.