Ethernet Glossary


A communications protocol developed by Apple Computer to allow networking between Macintoshes. All Macintosh computers have a LocalTalk port, running AppleTalk over a 230K bps serial line. AppleTalk also runs over Ethernet (EtherTalk) and Token Ring (TokenTalk) network media.


Clause 28 of the IEEE 802.3u standard specifies a MAC sublayer for the identification of the speed and duplex mode of connection being supported by a device. Support of this feature is optional for individual vendors.


Ability of a 10/100 Ethernet device to interpret the speed or duplex mode of the attached device and to adjust to that rate. Official term is Auto-Negotiation in Clause 28 of the IEEE 802.3u standard.


Attachment Unit Interface. A 15-pin shielded, twisted pair Ethernet cable used (optionally) to connect between network devices and a MAU.


Automatic determination and matching of transmission speed.


American Wire Gauge. A system that specifies wire size. The gauge varies inversely with the wire diameter size.


The main cable in a network.

Bandwidth on Demand:

Feature that allows a remote access device to initiate a second connection to a particular site. The network manager configuring the remote access server will specify a number of bits or a percentage of connection bandwidth threshold which will trigger the secondary connection. Multilink PPP is an emerging standard to allow this feature to be interoperable, but right now the only way to ensure correct operation is to use devices on both end from the same vendor.

Baseband LAN:

A LAN that uses a single carrier frequency over a single channel. Ethernet, Token Ring and Arcnet LANs use baseband transmission.


Unit of signal frequency in signals per second. Not synonymous with bits per second since signals can represent more than one bit. Baud equals bits per second only when the signal represents a single bit.


Binary, machine readable forms of programs that have been compiled or assembled. As opposed to Source language forms of programs.


Characteristic of having only two states, such as current on and current off. The binary number system uses only ones and zeros.


Specification for parallel printing which allows bidirectional communication on a Centronics-type interface. Pioneered by Hewlett-Packard, mainly used for postscript printers.


The smallest unit of data processing information. A bit (or binary digit) assumes the value of either 1 or 0.


A standardized connector used with Thinnet and coaxial cable.


A TCP/IP network protocol that lets network nodes request configuration information from a BOOTP "server" node.


Bits per second, units of transmission speed.


A networking device that connects two LANs and forwards or filters data packets between them, based on their destination addresses. Bridges operate at the data link level (or MAC-layer) of the OSI reference model, and are transparent to protocols and to higher level devices like routers.


A data transmission technique allowing multiple high-speed signals to share the bandwidth of a single cable via frequency division multiplexing.

Broadband Network:

A network that uses multiple carrier frequencies to transmit multiplexed signals on a single cable. Several networks may coexist on a single cable without interfering with one another.


A device that routes specific protocols, such as TCP/IP and IPX, and bridges other protocols, thereby combining the functions of both routers and bridges.


A LAN topology in which all the nodes are connected to a single cable. All nodes are considered equal and receive all transmissions on the medium.


A data unit of eight bits.


The data path between two nodes.


(Challenge Handshake Authentication Protocol) Authentication scheme for PPP where the password not only is required to begin connection but also is required during the connection - failure to provide correct password during either login or challenge mode will result in disconnect.

Coaxial Cable:

An electrical cable with a solid wire conductor at its center surrounded by insulating materials and an outer metal screen conductor with an axis of curvature coinciding with the inner conductor - hence "coaxial." Examples are standard CATV cable and Thinwire Ethernet cable.


The result of two network nodes transmitting on the same channel at the same time. The transmitted data is not usable.

Collision Detect:

A signal indicating that one or more stations are contending with the local station's transmission. The signal is sent by the Physical layer to the Data Link layer on an Ethernet/IEEE 802.3 node.

Communication Server:

A dedicated, standalone system that manages communications activities for other computers.


The terminal used to configure network devices at boot (start-up) time.


Noise passed between communications cables or device elements.


Technique for examining incoming packets whereby an Ethernet switch looks only at the first few bytes of a packet before forwarding or filtering it. This process is faster than looking at the whole packet, but it also allows some bad packets to be forwarded.


Carrier Sense Multiple Access with Collision Detection is the Ethernet media access method. All network devices contend equally for access to transmit. If a device detects another device's signal while it is transmitting, it aborts transmission and retries after a brief pause.

Data Link:

A logical connection between two nodes on the same circuit.

Data Link Layer:

Layer 2 of the seven-layer OSI reference model for communication between computers on networks. This layer defines protocols for data packets and how they are transmitted to and from each network device. It is a medium-independent, link-level communications facility on top of the Physical layer, and is divided into two sublayers: medium-access control (MAC) and logical-link control (LLC).


Digital Equipment Corporation (DEC) proprietary network architecture, a system for networking computers. It runs on point-to-point, X.25 and Ethernet networks.

Dial on Demand:

When a router detects the need to initiate a dial-up connection to a remote network, it does so automatically according to pre-defined parameters set by the network manager.


A security feature that ensures people do not log into modems that they shouldn't have access to. When a connection is requested, the system checks the user name for validity, then "dials back" the number associated with that user name.

Distributed Processing:

A system in which each computer or node in the network performs its own processing and manages some of its data while the network facilitates communications between the nodes.

Domain Name:

A domain name is a text name appended to a host name to form a unique host name across internets.


The transfer of a file or information from one network node to another. Generally refers to transferring a file from a "big" node, such as a computer, to a "small" node, such as a terminal server or printer.

End Node:

A node such as a PC that can only send and receive information for its own use. It cannot route and forward information to another node.


The most popular LAN technology in use today. The IEEE standard 802.3 defines the rules for configuring an Ethernet network. It is a 10 Mbps, CSMA/CD baseband network that runs over thin coax, thick coax, twisted pair or fiber optic cable.


Apple Computer's protocol for Ethernet transmissions.


Fiberoptic Data Distribution Interface. A cable interface capable of transmitting data at 100 Mbps. Originally specified for fiber lines, FDDI can also operate over twisted-pair cable for short distances.

Fiber-Optic Cable:

A transmission medium composed of a central glass optical fiber cable surrounded by cladding and an outer protective sheath. It transmits digital signals in the form of modulated light from a laser or LED (light-emitting diode).

File Server:

A computer that stores data for network users and provides network access to that data.


Process whereby an Ethernet switch or bridge reads the contents of a packet and then finds that the packet does not need to be forwarded and drops it. A filtering rate is the rate at which a device can receive packets and drop them without any loss of incoming packets or delay in processing.


Alterable programs in semipermanent storage, e.g., some type of flash reprogrammable memory.


Process whereby an Ethernet switch or bridge reads the contents of a packet and then passes that packet on to the appropriate attached segment. A forwarding rate is the time that it takes the device to execute all of the steps.

Flash ROM:

See ROM.


Dividing data for transmission into groups of bits, and adding a header and a check sequence to form a frame.


File Transfer Protocol, a TCP/IP protocol for file transfer.


Independent, simultaneous two-way transmission in both directions, as opposed to half-duplex transmission.


A device for interconnecting two or more dissimilar networks. It can translate all protocol levels from the Physical layer up through the Applications layer of the OSI model, and can therefore interconnect entities that differ in all details.

Hardware Address:

See Network Address.


The initial part of a data packet or frame containing identifying information such as the source of the data, its destination, and length.


Ethernet defined SQE signal quality test function.

Hertz (Hz):

A frequency unit equal to one cycle per second.


Generally a node on a network that can be used interactively, i.e., logged into, like a computer.

Host Table:

A list of TCP/IP hosts on the network along with their IP addresses.