Network Topologies

Your network topology defines the physical layout of your network cables as well as how your network clients physically connect to the network. There are three implementations in use in today’s networks: bus, star, and ring. As you will see, each topology has different strengths and weaknesses.

Bus

The low-end market utilizes a bus topology where each network adapter connects directly to the main network cable. Both thin Ethernet (COAX) and thick Ethernet are bus network design implementations. The basic layout of a bus network, as shown in Figure 8.4, is a single network cable connected from the first computer to the last in a single line. The first and last computer in the chain must be terminated. A terminator is nothing more than a resistor used to absorb the signal when it reaches the end of the network and prevent signal interference. Furthermore, one and only one end of the bus must be grounded to prevent ground loops.

Figure 8.4  A network using bus topology.

The bus topology’s primary flaw is that if one piece of the network cable is faulty, the entire network fails. And isolating the fault requires that you perform a binary test by splitting the bus into two separate sections to determine which half has the cable break. Once you have determined which half has the break, you must split that segment into two equal parts and continue this process until the cable break has been found. This can be a lengthy process. But don’t let this discourage you from using a bus topology-based network. It is still quite useful for setting up training rooms and small networks where the cables are easily accessible. In fact, I’ve set up several training rooms using this method.

Star

The majority of the network market utilizes the star topology. Star topology overcomes the single line break problem associated with bus topology by providing each computer with its own connection to a main network cable. Each workstation is connected to a multiport repeater, often referred to as a hub or concentrator. The hub, in turn, connects to the main network cable. The basic purpose of the hub is to retransmit the signals from the main network cable to the individual workstation, as shown in Figure 8.5. If a single cable fails between the hub and workstation, only that workstation is affected. The other workstations continue their network activity without experiencing any problems. However, if a hub fails, all the workstations connected to that hub fail. Of course, this is easily diagnosed because the users connected to the failed hub are clamoring for your attention. Unlike a broken cable in a bus topology which requires you to manually run around to each workstation in an attempt to isolate the broken cable, the star topology practically sends up a signal flare when a hub fails.

Figure 8.5  A network using star topology.

Planning for this type of disaster requires that you have a spare hub available to replace a failed hub, if such a problem occurs. This is one reason why it makes good sense to use the same type of hub throughout your network installation. In addition, it is a good decision to use smart hubs. A smart hub supports the Simple Network Management Protocol (SNMP). SNMP can be used to query a hub as to its performance and, in many cases, reboot it without physically walking over to the hub and cycling the power.

Ring

A ring topology is primarily utilized in IBM shops to support token ring networks. The physical layout appears identical to the star topology network displayed in Figure 8.5. Instead of a hub, the workstations connect to a Multiple Access Unit (MAU), which logically connects the network workstations into a ring, as shown in Figure 8.6. Ring topology offers the same benefits as star topology because the physical layout is the same. So, it follows that ring topology also suffers from the same problems as star topology, because the MAU is a central point of failure.

Figure 8.6  A network using ring topology.