Now that we've completed a general overview of the two types of networks you can build (peer-to-peer and server-based), we can discuss some of the basics related to the actual physical layout of LANs. Different topologies have been defined to characterize a LAN's layout (a topology is just how the network's cabling maps out).
Although these topologies do in some respects reflect the type of cabling (coaxial cable versus twisted-pair cable, both discussed in Chapter 4) that is used and the network architecture that is deployed (network architectures are also discussed in Chapter 4), they are just models and in many cases a LAN might use a hybrid of a couple different topologies.
A bus topology is characterized by a main trunk or backbone line with the networked computers attached at intervals along the trunk line, as shown in Figure 2.5. This topology type is considered a passive topology because the computers on a bus just sit and listen. When they "hear" data on the wire that belongs to them, they accept that data (they actually listen using their network interface cards). When they are ready to transmit, they make sure no one else on the bus is transmitting and then send their packets of information.
Figure 2.5. A bus topology provides one of the simplest ways to connect a group of computers.
Bus networks typically use coaxial networking cable (it looks like the same coaxial cable used for cable television, but it is actually slightly different) hooked to each computer using a T-connector. Each end of the network is terminated using a terminator specific to the cable type (if you use 50-Ohm cable, you use 50-Ohm terminators). Because the bus network is really just a collection of cable, connectors, and terminators, there is no amplification of the signal as it travels on the wire. This means that the size of the network will be limited by the maximum distance the cable type can actually move the signal that holds the data.
Bus networks are easy to assemble and are easy to extend. They require a fairly limited amount of cabling when compared to other topologies. Bus networks are prone to cable breaks, loose connectors, and cable shorts that can be very difficult to troubleshoot. One physical problem on the network, such as a detached connector, can actually bring down the entire bus network. Although at one time a bus network would have been the cheapest and easiest way to connect a small group of computers for peer-to-peer networking, the drop in the price of hubs (discussed in the next section on star networks) and the ease of using twisted-pair wire have really pushed the coaxial cable bus network to the edge of extinction .
The main reason for knowing about bus networks is that there are still bus installations found in small and medium-size companies. Remember that troubleshooting this type of network will typically require an inspection of all the cabling and their connections. As far as building a new network using a bus technology, you are far better off going with a star configuration using a hub (the star topology is discussed in the next section).
In a star topology , the computers on the network connect to a centralized connectivity device called a hub . Each computer is connected with its own cable (typically twisted-pair cable) to a port on the hub, as shown in Figure 2.6. Star LANs also use the same type of wait-and-listen strategy to access data or send data as characterized by the bus topology.
Figure 2.6. The star topology uses a hub as the central connection point for the computers on the LAN.
Because the star topology uses a separate cable connection for each computer on the network, star networks are easily expandable, with the main limiting factor being the number of ports available on the hub (although hubs can easily be daisy-chained together to increase the number of ports available). Expanding a star topology network is also very unobtrusive : Adding a computer to the network is just a matter of running a wire between the computer and the hub. Users on the network will be pretty much unaware that the expansion is taking place.
Disadvantages of the star topology revolve around cabling needs and the hub itself. Because each computer on the network requires a separate cable, cable costs will be higher than a bus topology network (although twisted-pair, the cable type used for star networks, is the least expensive cable type). Having to purchase a hub or hubs for your network does add additional costs when you are building a network based on the star topology, but considering the benefits of this type of topology in terms of managing the physical aspects of your network, it is probably well worth it.
The greatest benefit of using the star topology is that you can easily add new computers to the network without disrupting service to the computers already on the network. Also, if one computer goes down on the network, it does not negate the ability of the other computers on the star to communicate with each other. Obviously, the most crucial failure point on a star network would be the central hub.
A ring topology connects the LAN computers one after the other on the wire in a physical circle, as shown in Figure 2.7. The ring topology moves information on the wire in one direction and is considered an active topology. Computers on the network actually retransmit the packets ( packet is a generic term for the chunks of data that are being moved along the network) they receive and then send them on to the next computer in the ring.
Figure 2.7. The ring topology connects the computers in a circle, where a token is used to permit the transmission of data.
The ring topology is considered an active topology because the computers in the ring actually pass a token around the circle. The token is a special packet of data, sort of like your ATM card, that gives the computer with the token special abilities . If a computer wants to send data onto the network, it has to wait until it has possession of the token. This is how the IBM-Token Ring network architecture operates (which is discussed in Chapter 4).
True ring topologies can be difficult to troubleshoot, and the failure of one computer on the ring can disrupt the data flow (because data circulates around the ring in one direction). Also, adding or removing computers from this type of topology can disrupt the operation of the network. On most small LANs, you won't run into the ring topology because IBM Token Ring and some other high-speed network technologies that use a ring are more typical of larger networks.
Since Token Ring hardware is expensive and Token Ring networks require a substantial knowledge base to administer, ring topologies are only used on larger corporate networks. You would not use this type of topology for a home or small office network. The star topology is the cheapest and easiest to deploy in a small office situation.