Published In July 2001
A recent survey of network executives published in Network World revealed the “typical” business network architecture today. It was interesting to learn that about three-quarters of local area network (LAN) cabling today is Category 5 and less than 20 percent is more advanced than Category 5, including Category 5e and Category 6. A more surprising result from the same survey is that almost half of the LANs today use switched Fast Ethernet (100Mbps) and the remainder is divided roughly between shared 10Mbps (which I call classic Ethernet) and switched 10 Mbps. Only 1 percent of the LANs surveyed used Gigabit Ethernet today (exclusively), but as I have previously discussed in this series, GigE is being introduced and so far mainly restricted to the core or backbone portion of today’s enterprise network or intranet. What else is happening in the world of Gigabit Ethernet? I’ll highlight several other ongoing developments in this article. Power in the data cable is coming back. Many, if not most, LAN cables used today are purely data, with no power conductors. The last time I saw power running in the data cable was in the early 1980s: a Private Branch Exchange (PBX) needed its telephones and computers to share a central power source and ground. This meant there would be no need to plug a telephone into power separately and also that there would be no current loops to cause noise problems in the system. In that case, there were only two pairs in the four-pair cable used to carry digital voice, meaning that the other two were available. (The fourth pair was used to carry a traditional analog talking circuit.) And what is driving this need for power in a data cable? Perhaps one of the strongest motivations to power devices through the same single cable is voice over internet protocol (VoIP), which is rapidly becoming popular in business environments. (One example is Cisco Systems, which is in the process of replacing its own internal traditional voice communications systems with new systems based upon IP telephony.) One alternative would be battery power in the telephone that connects to the Ethernet, but this can create enormous administration, management, and inventory problems. After all, the telephone switch supplies power to telephones, so why shouldn’t its successor, the data switch, also supply telephone power? Other types of devices that might take advantage of power delivered by data cable would include burglar and fire alarms, thermostats, or any other smart appliances. In a twist on residential intercoms that carry the voice signals on power wiring, intercom products might use Ethernet cable with power included. Today the Institute of Electrical and Electronic Engineers (IEEE) 802 project has launched the IEEE 802.3af task force and given it the mission of providing a standard way to power a device via a data cable. Note that the parent IEEE 802.3 committee is the one that standardizes Ethernet, Fast Ethernet, and Gigabit Ethernet. Although there is no specification of speed associated with this task force, we expect that the standard will apply to all speeds: 10, 100, 1,000, and soon, 10,000Mbps. Another issue that needs to be considered is whether the cabling is Unshielded Twisted Pair (UTP) or fiber optic. Although the 802.3af effort specifically targets Category 5 UTP cable, the standard should at least accommodate the same type pairs incorporated into the fiber cable to make the hybrid wiring closet easier to work in. An even more difficult concern is standardizing power in cables for other than Ethernet. What about fiber distributed data interface (FDDI), asynchronous transfer mode (ATM), and even dense wave division multiplexing (DWDM) cables? When do these diverse words converge? All involve Ethernet, of course! In our discussions so far, we have mentioned primarily business applications except for residential examples in the data cable power section above. But more and more homes and businesses are being wired with data circuits and power and telephone. Is it possible that these will all converge into single power/data cables in the near term? Apart from the vendor and union issues that they may create, what technical issues will there be? Safety? Electromagnetic interference (EMI)? Minimum-bend radius? Insulation? Longevity? Another emerging concern about Ethernet is operations, administration, maintenance, and provisioning, or OAM&P. This topic has long been considered crucial in the telephone and cable television businesses, but has been minimized in business LANs. Now, however, a remote office or a residence often contains Ethernet, and while it is simple, it is not so easy for a technician to do a hands-on test of that network. Furthermore, the revenue stream coming from the Internet-enabled services associated with such Ethernets will not support frequent technician site visits for testing or configuration. Hence, we should expect more self-testing and remote testing to be developed for Ethernets. At the same time that Ethernet seems to be nearing a total takeover of the LAN market, it is being used for more diverse purposes. For example, it was recently reported that the new Tokyo DisneySea theme park will use exclusively Fast Ethernet. At first, this is surprising because up to 40 channels of audio will be supported on the system. Audio usually is considered difficult to handle on a LAN because it has no guarantee that sound will be delivered without jitter (delay variations). But with the high capacity of Fast (and if needed later, Gigabit) Ethernet, it was chosen over ATM, which some might consider the natural choice. SHINN is principal of Compututor. He can be reached via e-mail at firstname.lastname@example.org.