This chapter discusses Local Area Networks, the most common kind found in office settings. The objectives important to this chapter are:
Concepts:A Local Area Network (LAN) is the smallest of the three kinds of networks defined in chapter 11 (WAN, MAN and LAN). Typically, a LAN is owned by the organization that uses it, it is relatively smaller and simpler than the other two types of networks, and is may provide users with services at higher speeds than are attainable over larger networks. Depending on the type of LAN, the size could be a room or several buildings, the ownership is typically private (you can set one up at home for practice) and the speed can vary from 2 to 100 Mbps.
Common reasons for creating a LAN include sharing resources and information. It is more efficient for a group of people to share printers on a LAN than to buy a printer for each user. It is possible to make sure users are all running the same version of a program if they are running it on a server on the network. It is possible to share sensitive files and email on a LAN that you would never want shared on a network that used a PDN connection.
Some applications are written to take advantage of networks, and can provide cost savings. A site license is an agreement with a publisher to pay for the use of software either run on a network or installed from a network. The buyer pays for a certain number of users, and the price is usually less than buying the same number of copies of the software for that many individuals.
The text mentions that LANs caused the need to expand the OSI model. On page 455, you see that the Data Link Layer is divided into two sublayers. The Logical Link Control (LLC) sublayer carries out the frame functions, but the Media Access Control (MAC) sublayer specifies how a machine gets access to the network medium.
Before discussing the LAN topologies, remember that each kind of LAN can have two topologies: physical and logical. The physical topology describes how the LAN is actually wired, while the logical topology describes how it works.
Two types of cables are mentioned on page 457. Many LANs use either UTP or coaxial cable to connect their nodes. UTP is cheaper, but coaxial cable has greater bandwidth. Coaxial cable can have greater speed, but this is not always true.
A hub is also called a concentrator, because it gathers a number of wires in one location (concentrates them). Hubs are often used in UTP LANs, and can come in several types. Two are described: repeaters, which regenerate signals; error correcting hubs, which try to improve signals.
Wiring cost is discussed on page 458. Better wire costs more, but as a general rule, buy a better product than you need today. You may wish you had bought it tomorrow.
Wireless LANs are mentioned briefly. Some networks are "wireless" in that each node is meant to be mobile and connects to the network with infrared or radio signal. Cellular phones work this way, although they are often actually MANs.
The concept of transmission starts on page 459. Two techniques are discussed:
The chart on page 461 shows several possible ways a LAN can be implemented, all of which follow the IEEE 802.2 standard for their LLC sublayer. The variety comes in the wiring method and access method used. The chart on page 462 shows four common topologies and compares them on five characteristics: transmission, access method, topology, medium and speed. Notice that all four LAN types shown use baseband transmission. Ethernet uses CSMA/CD and Token Ring LANs use Token passing for their access methods, while Appletalk (an Apple standard) and Arcnet (an ancient standard) are listed as "nonstandard", which means proprietary access methods are used. The actual physical topology used for these networks varies as does the medium, but most are commonly wired as stars, usingcoaxial cable or UTP. Speeds vary as shown, except that Arcnet is commonly 2.5 Mbps, not 25.
Each LAN type is discussed in greater detail. The chart on page 464 shows the characteristics of eight varieties of Ethernet, another product invented as Xerox and made famous by other vendors. The chart corrects an error in the text. 10BASE2 LANs have a maximum segment length of 185 meters, not 200 meters. A segment could be defined as a number of nodes connected by one hub (if using UTP) or the nodes connected to one continuous length of coaxial cable, if using that kind of wire. The most common type from this chart is 10BaseT.
Token Ring technology was invented by IBM. It can be wired as a star or a ring, but its logical topology is always a ring. Speeds are commonly 4 or 16 Mbps.
FDDI is an interesting standard, discussed on page 467. It is actually two token rings. Workstations can be on one or both rings, which are typically "wired" with fiber optic cable. Class A workstations are on both rings, Class B workstations are only on one ring. One ring is sufficient for the LAN to work, but the second ring gives stability. If there is a break in the cable of a single ring Token Ring LAN, the whole LAN has a problem, but the second ring in FDDI allows the two stations that are cut off from one another to "wrap" the ring. They pass messages back along the second channel, making the ring longer, but still intact.
The AppleTalk topology is discussed on pages 468 and 469. It is described as a proprietary form of CSMA/CA. Collisions are avoided by having each sender send out a small frame that warns other stations of its intent to transmit a message. Stations seeing the warning frame stay off the LAN until the message passes. Of course, collisions of these frames are still possible.
Arcnet is discussed on page 459. It is an older standard, usually very slow by LAN standards at only 2.5 Mbps. Being an older standard, it does not interface well with other standards developed since it was created.
Five types of communications devices are described on the following
On most networks, the client-server model is used. Workstations are usually clients that receive services from more powerful computers called servers. Several kinds of services are possible, including print service, file service, communications service and application service.
To put a computer on a network, you usually have to make sure the computer has a Network Interface Card (NIC), that must match the kind of computer you put it in, the kind of network and the kind of cabling in use. In addition, the workstations and the servers must run some matching network software, or they will not communicate. Four major brands of network software are listed on page 484: NetWare (from Novell), Windows NT (from Microsoft), LAN Server (from IBM) and VINES (from Banyan). To learn to manage any one of these properly will take several courses.
Several criteria are discussed on pages 485 and 486 for evaluating the
performance of a LAN. This is done to determine if the LAN needs
work, improvement or replacement.
On page 492, there is a discussion of several ideas about managing a
LAN. The topics are not all obvious to an outsider to the business. Some
important ones are: