Those who write such network standards as Ethernet have a broader charter. They have to consider not only the “physical layer,” as they refer to the cabling and transceivers (transmitters and receivers of the network hardware), but also the network transmission protocols that carry encoded data signals, message addressing and allow sharing of the network bandwidth. Some of these standards will depend on the cabling meeting certain minimum standards to allow networks to operate over specified distances with virtually 100 percent certainty.

Obviously, the two groups must communicate during the standards-making process, but their goals are different. Both are composed of companies interested in selling products in the respective markets. Network standards groups are interested in those that support users' demands for higher speeds so their market is continually expanding. Cabling standards groups are interested in creating new cabling to support higher-speed networks that will result in new installations to expand their market. Both groups can meet their needs-expanding their market-while fulfilling the demands of users. Everybody wins.

Sometimes in the process, the results of each group conflict. For example, when network speeds get higher, the technical demands on cabling become greater. Cabling manufacturers respond by increasing the performance of their cabling, while the network manufacturers try to develop equipment that will work on the current generation of cabling. That's how we get into situations where a lot of Gigabit Ethernet equipment has been tweaked to work on users' currently installed Category 5 UTP, while cabling manufacturers have been pushing enhanced Category 5 (Category 5e) and Category 6 for Gigabit applications. The jockeying over which UTP cabling will work with 10-Gigabit Ethernet is creating a similar situation.

There is a similar situation with fiber optics. Basically all applications of local area network (LAN) cabling in the United States use a 62.5/125 micron multimode fiber with bandwidth performance specified for FDDI-a 100Mb/s network introduced 15 years ago. It is generally referred to as FDDI fiber. As with all fiber-versus-copper comparisons, FDDI fiber had big advantages over UTP or coax cable, allowing FDDI and Fast Ethernet to run up to two km, fully 20 times further than Category 5 UTP. FDDI-grade fiber has supported the expansion of networks like Ethernet from 10 Mb/s to 10 Gb/s-a thousandfold increase in speed-simply by reducing the distance limit.

In order to reach fiber optic network speeds, transceivers have changed from LEDs to lasers, since LEDs top out at about 200 to 300 Mb/s. Gigabit network transceivers use a combination of 850 nm VCSELs (vertical cavity surface-emitting lasers) and 1,310 nm fabry-perot (F-P) lasers that can be modulated at higher speeds and have higher effective bandwidth in the fiber. Just like with copper, network equipment manufacturers have been able to use the currently installed FDDI fiber cable plant for faster networks by crafty equipment design.

Eventually, of course, you reach a “brick wall,” and fiber optic networks are approaching that point. FDDI fiber will support 10 Gigabit Ethernet only to about 30 meters, long enough for a computer room but too short for a backbone. Thus the standards for 10 Gigabit Ethernet call for alternative fiber types, either of two types of 50/125 micron multimode or single-mode fiber.

Single-mode fiber has been used in LAN backbones for ages, as it allows virtually unlimited lengths. However, experience has shown that care must be taken with single-mode terminations, as back reflections at connectors can produce multiple reflections in short cables, creating background noise that adversely affects receiver performance. Basically, single-mode cabling should only use high-quality preterminated pigtails fusion spliced onto the cable for terminations, a fairly expensive process.

The alternative of laser-optimized (LO) 50/125 micron fiber allows up to 300 meters backbone distance with more conventional installation and termination procedures. The lower-grade 50 micron fiber (500 MHz-km bandwidth) allows about the same distances as structured cabling calls for and actually costs less than FDDI grade fiber. The higher spec fiber (2,000 MHz-km bandwidth) costs a little bit more but provides significantly more distance capability. And any network that currently uses FDDI fiber will work over the LO fiber. Standards are including LO fiber as they are revised.

New fiber optic networks should use LO fiber if upgrades to 10 Gigabit Ethernet are contemplated, which it probably will be. Users are often reluctant to make the conversion if they already have the 62.5/125 FDDI fiber installed, as the two fibers cannot be mixed in one link due to high losses caused by the different core sizes. However, proper documentation and color coding of each fiber type can be a solution. EC

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