Published In June 2001
Recently, there has been a surge of interest and activity in using Ethernet as a wide area network (WAN) technology. This is surprising because Ethernet was developed as a local area network (LAN) technology, serving to deliver data between computers within the same building. This column examines its attraction for WANs. How can we use Ethernet on WANs? To be precise, modern WANs only use part of Ethernet—its frame format is borrowed but not its Carrier Sense Multiple Access/Collision Detection (CSMA/CD) function. In other words, WANs use the same data organization, but not the scheme for accessing the electrical bus. This is appropriate because there’s no bus: most WANs now operate on optical fiber cables with only one device on each end. Hence, there’s no need to arbitrate access to the cable because the two devices (usually routers) on the ends of the cable can both send simultaneously in full duplex mode. There can never be a collision on an optical fiber strand with all transmissions going in the same direction. Ethernet frames are formatted according to this sequence. Note the numbers above the fields, which label their lengths in bytes, or octets, as the network standards call them. The advantage of this frame organization is that it is simple and yet expresses what is needed to handle the data, either on the LAN or on the WAN. It has minimum overhead (extra bytes). The routers that take the frame off the LAN can easily use the same type of frame to put it onto the fiber optic WAN. Metropolitan fiber data networks have been using this approach for years. In fact, MFS (now part of WorldCom) created the metropolitan area Ethernet (MAE-East) in the Washington, D.C., area which is now one of the largest network access points (NAPs) on the Internet. Recently, many firms have been building metropolitan area networks based upon Gigabit Ethernet using the frame illustrated. One of the leaders in this market is Yipes Communications. Yipes claims it can offer from 1 Mbps to 1,000 Mbps at less than half the price of a comparable service from a traditional carrier. (Traditional carriers would typically supply this need with T-1 or T-3 lines.) A number of other firms are also looking at running Gigabit Ethernet in the wide area. In fact, the IEEE standards group is working to develop a standard so far tentatively called Ethernet in the First Mile (EFM). These developments are especially interesting in view of the historical battle for access market share. Access is the first mile (or the last, depending on where you start) connecting the customer premises to the network carrier’s infrastructure. When asynchronous transfer mode (ATM) started, it was a WAN technology. Later, some vendors, notably IBM, attempted to adapt it for use as a LAN. Some feared that this effort, dubbed “ATM to the Desktop,” would eventually supplant Ethernet. Instead, ATM/Desktop was found to be too expensive and complex and we now find that Ethernet, originally a LAN technology, is also spreading its wings as a WAN! A major reason for the proliferation of Ethernet into the WAN is the use of (fiber) glass cables. Such cables offer enough bandwidth to easily carry Gigabit Ethernet and furthermore are ultra-reliable when compared to any metal cables. Thus, other more complicated schemes for support of metal cables are not needed and the elegant but simple Ethernet frame suffices to carry any kind of information over the fiber cables. At this moment the access market seems to be saying that more bandwidth is the cure, not more complicated means of ensuring that voice, data, and video are properly prioritized. This argument has been ongoing for years, but today it is said that there is a glut of long-haul fiber available, thanks to dense wave division multiplexing (DWDM) and extensive implementation programs. This may well lead to a rapid buildout of short-haul connections to reach and take advantage of the glut of long-distance circuits. How does all this information impact electrical contractors? I can think of two ways. First, the struggle for higher-speed access continues, even in the face of the high-tech business crisis. Some are predicting that Gigabit Ethernet (Ethernet in the First Mile or EFM) might displace both Digital Subscriber Line (DSL) and cable modem mechanisms for broadband access, perhaps even in the next two to three years. So contractors should be alert for continuing, perhaps even expanding, opportunities to wire (or re-wire) neighborhoods for GigE, starting with businesses and continuing to residences. Second, even if the promising economies of scale and ease of use fail to materialize and fail to deliver that first mile to Gigabit Ethernet, it seems clear that the inside wiring itself will have to be revamped for high-speed interconnections. Thus, our growing set of electronic gadgets will demand richer connections. Even if some of them are wireless, they’ll need access to our wired networks, and contractors will need to connect their antennas! Next month, I’ll discuss a new initiative (and a new reason) to mix electrical power wiring with data wiring within the same cable. SHINN is principal of Compututor. He can be reached via e-mail at firstname.lastname@example.org.