This article inaugurated the Inside Technology feature in the January 1997 Hawaii Library Association Newsletter.

What is Ethernet?

Back in the late 1600s, physicists trying to explain the behavior of light theorized that light was carried by an undetectable substance called ether (not to be confused with the anesthetic) that filled all space. That theory was disproved by the late 1800s, but the figurative use of the word in describing the passage of radio waves through the heavens remained. So when Xerox developed a networking system that allowed computer devices to communicate with each other using radio-like signals over antenna cable in the early 1970s, they called it Ethernet.

Ethernet is one of the most popular systems for physically connecting computer devices together to form a local area network, or LAN. Other systems that perform the same function using different methods include ARCnet and Token Ring.

Ethernet manages the transfer of data between computer devices, and facilitates higher level networking systems like Novell Netware, AppleTalk (using Apple's EtherTalk), Microsoft Windows Network, Banyan VINES, and TCP/IP (the language of the Internet). These systems govern how data and computing resources (like printing and hard disk storage) are actually shared.

Ethernet operates on a model known as Carrier Sense Multiple Access with Collision Detection, or CSMA/CD. Each device on the network is connected to a common communications channel, and any device can use the channel at any time. Each device listens to the channel for data signals which may be transmitted by other devices. If one device has data to send to another device and the channel is clear, it will transmit the data in a special format called an Ethernet frame. If the channel is busy, the device waits until it is clear before trying to transmit. If by chance, two or more devices try to transmit at the same moment, the signals will collide. The collision is detected, everything stops, and after a seemingly random waiting period, one of the devices will try to transmit again. If necessary, the process will repeat until one device has uncontended use of the channel. This system works well at low-to-moderate activity levels. Ethernet's efficiency, simplicity, and flexibility make it a popular choice for applications that aren't highly active and don't have a critical need to pass data in a timely manner.

To help visualize how this works, imagine a room occupied by a group of people, all within earshot of each other. If one person needs to tell another person something, he or she will wait until nobody else is talking, and then say, "Hello F3AC567B3DF2, this is F3D92A50FCAB, my message is . . ." But if two people try to talk at the same time, they will hear each other, stop talking, and eventually one person will try to speak again. The content and the language of the message itself could be anything -- they may be speaking in Novell and talking about transferring a file from one computer to another. Ethernet only governs how the conversation is conducted.

Here's a bit of trivia for you -- A radio-based CSMA/CD system called ALOHANET was developed at the University of Hawaii, and predates Ethernet by several years.

The typical device on an Ethernet LAN is a microcomputer with an Ethernet adapter card installed. Other devices include print servers, which provide centralized printing (this could be a specialized device or a microcomputer acting as a print server), terminal servers, which allow ASCII terminals to use the network to reach remote hosts (OPAC terminals in many libraries are connected this way), or minicomputer or mainframe host computers. Specialized network devices include repeaters, which connect network segments to extend a LAN's range or connect different cable types, bridges, which connect different parts of a larger network and can join different kinds of LANs, and routers, which determine the best way to transfer data throughout a larger network.

An Ethernet LAN can take several forms. Two forms use coaxial cable in what is called a bus topology, in which all devices are attached to a single long cable. Standard Ethernet, also known as thick net, uses cable that's about a half-inch thick. It is usually used to span long distances, such as for connecting several buildings on a campus-wide network. A device called a transceiver is attached to the coaxial cable and a separate data cable runs between it and the Ethernet card. It is also known as 10BASE5, because it transfers data at a rate of 10 million bits per second, uses baseband transmission (using the whole capacity of the channel for a single signal), and can be up to 500 meters long. 10BASE2 (or ThinNet, thin net, or cheapernet) uses cable that's about a quarter-inch thick, and can be up to 185 meters long. The bus cable is connected to the Ethernet card, which has the transceiver built in. It can be a cost-effective option for small LANs. 10BASE-T uses inexpensive unshielded twisted-pair cabling (similar to telephone cable) in a star topology, in which each device has its own cable (up to 100 meters long) connected to a central device called a hub or a concentrator. 10BASE-T is extremely popular in larger LANs because it is flexible and easy to troubleshoot because each individual device can be isolated. Other implementations include 10BASE-FL, which uses fiber optic cables and is good for up to 2 kilometers, and 100BASE-T Fast Ethernet, which has a data rate of 100 megabits per second.


Sources: Inside TCP/IP, Matthew Flint Arnett et. al., New Riders Publishing 1994; NetWare Training Guide: Networking Technologies 2nd ed., Debra R. Niedermiller-Chaffins et. al., New Riders Publishing 1994; Black Box Catalog, Summer 1996; World Book Encyclopedia (Ether) 1976.