Local-Area Network (Lan) | Research & Encyclopedia Articles

Local-Area Network (Lan)

The term "local-area network" (or LAN) describes a group of computers connected by usually no more than 1,000 feet of cable that allows these computers to share data. LANs can be created using different types of cables, configurations, and protocols, and still remain within the requirements of a network system specification.

Networks consist of four basic components:

• Network Operating System (NOS). The NOS consists of several computer programs that run in the networked computers. Some of these computers share files, printers, and other peripherals over a network and are called servers. Other computers that use these shared resources are called clients.
• Network Peripherals. Printers and scanners with their own network connections are examples of network peripherals. These components use specialized processors to operate networking server software without having to be connected to a computer.
• Network Interface Card (NIC). NICs are also called LAN adapters. NICs have several functions inside a computer. They convert the low-power signals from inside a computer into stronger signals that can propagate across a network cable. The NIC also prepares data for transmission by including important messaging or packaging information that is used by other computers on the LAN. The NIC also controls access to the network cable that is shared by other computers on the network.
• Network Cabling. Network data can be sent as 1) electrical pulses over different types of copper cable, 2) pulses of light over fiber optic cable, 3) through the air as radio or light waves, or 4) data can be sent via a combination of these.

Major Types of LAN Technology

Asynchronous Transfer Mode (ATM)

Asynchronous Transfer Mode (ATM) is a connection-based type of networking that can be used in LANs and wide-area networks (WANs). ATM is what is responsible for high-speed data switching, often at gigabit rates. However, most computers have a difficult time being able to absorb data at gigabit speeds. ATM networks, therefore, handle traffic from many other computers and are more expensive to build and maintain.

ATM networks use optical fibers for connections because optical fibers transfer data at a higher rate than does copper (ATMs can operate at speeds between 1.544 Mbps to 1.2 Gbps). The lowest levels of an ATM network use fixed-size data packets called cells. The one-size cells allows ATM switch hardware to process them quickly.

Because ATM networks carry voice and data, many telephone system networks use ATM technology.

Bridge

A bridge is a computer that connects two or more networks and forwards data packets among them. A bridge can also link the faster portion of a LAN to slower links, such as networks of leased telephone lines. Bridges help extend the network and help isolate network traffic. Bridges send packets (small blocks of data sent across packet-switching networks) and frames (packages of data) between different types of media, but they can only forward information that is addressed to devices on another network system.

Local bridges connect fast cable segments on one LAN (this usually requires one "local" bridge). Remote bridges connect fast local cables to slower, longer-distance cables in order to connect physically separated networks (this usually requires two "remote" bridges).

Broadcasting

In LAN context, a broadcast is a packet-delivery system that delivers a copy of a particular packet to all hosts that are attached to it, much the way a television signal is broadcast to all the television stations tuned to it. Broadcasting is a technique used by the most popular network connection technology, Ethernet.

With a broadcast-network system, a single node on the network will transmit information to all other nodes at the same time. However, not all of the nodes will receive the message; only those nodes that have been addressed in the message will be able to receive it. This is similar to the television signal analogy where, for example, only those televisions able to receive an HBO or pay-per-view signal (because a signal blocking filter has been removed) can receive the broadcast. In the LAN world, once the signal has been received, the receiving stations will send an acknowledgement to the sending station.

Carrier-Sense Multiple Access (CSMA)

Carrier-Sense Multiple Access (CSMA) is a method used in Ethernet networks to transmit data only during breaks in the network traffic. This method is typically used by network interface cards that share a common cable. CSMA with collision detection, also known as CSMA/CD, is a technique used if two transmit stations begin transmission at the same time, they can detect a collision, stop, and retry after a predefined period of time. Ethernet networks use a binary exponential backoff policy, which is simply an algorithm that helps avoid massive data collisions in the unlikely event many transmitting stations try to transmit simultaneously.

Ethernet

In the late 1970s IBM and Digital Equipment Corporation developed methods for several large computers to operate over local networks. Later, Xerox Corporation developed a method that allowed many computers to be connected over a LAN. This method later developed into a set of standards (minimum performance requirements) and protocols (rules for how different elements interact) called Ethernet, which today is the most popular type of LAN.

Today's Ethernet LANs allow for mixing and matching of various hardware and software, network cables, and configurations, and still remain within the boundaries of the Ethernet specification, also known as IEEE 802.3. Today the Ethernet can achieve data-transfer rates in excess of one Gigabit (one Gigabit = 1,000 Megabits).

In thin Ethernet networks, the cable configuration uses a T-shaped coaxial connector at each network interface card (NIC). The more common Ethernet configuration uses unshielded twisted-pair wire available in three configurations: 10Base-T (10 megabits per second, or Mbps, signal speed), 100Base-T (100 Mbps, also known as fast Ethernet), and 1000Base-T (1000 Mbps, also known as Gigabit Ethernet). Unshielded twisted pair wire uses a small plastic connector known as an RJ-45 connector at each end of the wire, which plugs into the back of a computer, wiring hub, or NIC.

Fiber Distributed Data Interface (FDDI)

Fiber Distributed Data Interface (FDDI) is a type of high-speed LAN that uses optical fiber to carry signals that contain data encoded in pulses of light. Another similar technology called Copper Distributed Data Interface (CDDI) works similar to FDDI, but uses copper cables to carry signals.

Unlike other copper-based networks, optical fiber is immune from electrical interference, and because optical fiber uses light, it can transmit much higher data rates than copper cables carrying electrical signals (per unit time). FDDI is known as a "ring" because the network configuration starts and ends at the same source. The most unusual fact about FDDI is its ability to detect and diagnose problems; in fact, because the network hardware can automatically accommodate a network failure, it is sometimes called a self-healing network.

IBM Token-Ring

Token-Ring network interface cards use a complex network cable-access control methodology. While Ethernet NICs contend for cable access, Token-Ring NICs require permission to transmit into a cable that forms a completed loop. The active Token-Ring NICs "negotiate" (using their serial numbers) to determine which card is the master NIC. The master NIC transmits a special message called a "free token." When a data-loaded NIC receives a free token, it converts the free token into a message and forwards it to the next station on the loop for relay. After the addressed NIC receives the message, and the message returns to the originating NIC, that master NIC generates another free token and the process repeats.

A Token-Ring network has two physically separate wiring hubs that are connected by fiber optic cable that can be thousands of feet apart. However, the devices attached to the Token-Ring network (connected by unshielded twisted pair, or UTP, wires) must be no farther than 100 feet from the wiring hub. The wiring configuration more resembles a star-shape, whereas the actual "ring" in the Token-Ring network is contained within the wiring hubs.

Multicasting

Broadcasting involves the delivery of a packet to all computers on a specific network. Unicasting involves sending a packet to a single device. Multicasting involves the delivery of a packet to a specific subset of hosts. Unlike broadcasting, multicasting allows each system the choice of participating in a particular multicast.

Large numbers of addresses are reserved in a hardware device that will later be used in the multicast. When a collection of devices wants to communicate, one specific multicast address is selected for use. All connected devices will receive a copy of any packet sent to the multicast address, once the NICs have been configured to recognize the selected multicast address.

A type of multicasting that was popular for a short time in the late 1990s was called Push. Push (or "server-push") is the delivery of information on a network (in this case, the World Wide Web) that is initiated by the information server rather than by the information user or client, as it usually is. An early Web service that specialized in "pushing" information rather than having it "pulled" as the result of requests for Web pages was Pointcast, a site that provided late-breaking news and other information customized to a previously defined user profile. Marimba was another similar site (and product) that pushed information to the user on a predefined schedule.

The information pushed from a server to a user actually arrives as the result of a programmed request from the client in a computer. Any information pusher on the Web requires the download of a client program. This program captures a user profile and then occasionally initiates requests for information on the user's behalf from the server.

LAN Topologies

Topology is simply the physical layout and configuration of the devices on a network, or the "network architecture." There are four fundamental LAN topologies.

• Star Topology. The star topology is the oldest type of network architecture. It is similar to a telephone system in which calls from one person to another are handled through a central switching station. With star LANs, all messages are routed through a central hub. Adding nodes to a star topology is simple and easy: just add a NIC and cable to connect to the central hub. This configuration allows certain nodes to be given higher priority over others. The central hub simply searches for signals from these higher priority nodes before acknowledging the others.
• Clustered Star Topology. A clustered star topology is simply several star topologies linked together. In this configuration, if one star malfunctions or fails, the entire network does not fail.
• Bus Topology. In this simple configuration, a bus (single master cable carrying all traffic) serves as the focal point for the LAN, where the nodes connect along the length of the bus. Devices on the network need to be sufficiently spaced apart to avoid electrical interference. The problem with such a configuration is that one malfunctioning node on the bus can shut down the LAN.
• Mesh Topology. A mesh topology is exactly what it sounds like. It is a network without a defined network architecture or topology. Such topologies tend to be extremely complex and have all types of built-in redundancies to minimize network downtime or failures.

This is the complete article, containing 1,811 words (approx. 6 pages at 300 words per page).