The information in this visual guide, based on definitions in the upcoming Network+ Exam Cram 4th Edition, illustrates the various physical network topologies you need to know for the soon-to-be-released fifth edition of that certification exam.

To make the concepts easily understandable, imagine the analogy of a hallway in a campus building. Faculty offices are on both sides of the hallway and at the end of the hall, an administrative assistant sits outside the dean's office.

To further the analogy, assume that one of the faculty members -- Dulaney -- needs to send a message to the dean telling him that 17 students are currently enrolled in the e-business course. The various physical topologies will indicate the method by which that message can go from sender to receiver.

Ring Topology

In this configuration, there is a student assistant who makes a loop through the offices once each day, stops at each door and stops the gum chewing just long enough to ask, "Do you need anything done?" If I happen to be paying attention when this barely audible mumbling occurs at the door, I can respond, "Yes -- tell the dean we have 17 students in e-business," and trust the message will be delivered. If I am not paying attention, I'll miss the mumbling altogether and have to wait until she makes another round (probably tomorrow) before passing along the message.

Since the student can't hold too much information, once I load her brain with my message, she will fail to hear any that may be given to her by Merle, Seaborne or Smith and not be able to think about anything else until she drops the message with Truitt. The horrible part of it is that Buck -- at whose office she stops right before arriving at my door -- may be giving her hundreds of messages to deliver, and her brain is already full before she ever gets to my door. 

The ring topology in the real world.

The ring topology truly is a logical ring -- data travels in a circular fashion from one computer to another on the network -- and neither a hub nor switch is needed. An electronic token must be captured in order to send data, and if a single computer or section of cable fails, the signal is interrupted and the entire network becomes inaccessible. While ring networks are moderately easy to install, expansion can cause the network disruption and they tend to work better with smaller networks.

Bus Topology
In this configuration, I just stand up and yell at the top of my lungs, "We have 17 students in e-business!" Those other than the dean should ignore the message, while he listens to it. Because I am quite a yeller, the two ends of the hall have soundproofing added to them to absorb what they encounter or else my shrieks would bounce throughout the building and interrupt classrooms and others, not to mention reverberate up and down the hall a few times, making other conversation impossible. If Merle decides to add Styrofoam to her walls, the whole system falls apart.

The bus topology in the real world.

A bus topology uses a trunk -- or backbone -- to connect all the computers on the network, and individual systems connect to this backbone using T connectors or taps. To avoid signal reflection, a physical bus topology requires that each end of the physical bus be terminated, with one end also being grounded. You don't need a hub or switch, making this fairly cheap to implement. When a computer is added or removed, a break in the cable prevents all systems from accessing the network.

Star Topology

In this configuration, I address my message to the dean, put it in a campus mail envelope and let it get picked up. Every campus mail envelope goes through Smith, the administrative assistant. She decides whether it gets routed to the dean, or to another faculty member. Nothing happens in the hallway that she is not aware of and approving of; if anyone tries to send junk mail, she discards it so fast your head will spin.

The star topology in the real world.

In the star topology, all computers and other network devices connect to a central device (a hub or switch). Each connected device requires a single cable to be connected to the hub, creating a point-to-point connection between the device and the hub. Using a separate cable to connect to the hub allows the network to be expanded without disruption. A break in any single cable does not cause the entire network to fail and it becomes the easiest to expand and the most widely implemented network design in use today.

Because all devices connect to a centralized hub, this creates a single point of failure for the network (what if Smith calls in sick?). If the hub fails, any device connected to it cannot access the network. It also tends to require more cabling than a bus or ring and requires the additional equipment cost.

Mesh Topology
In this configuration, there is a direct connection between every office. If I jump up and run to tell the dean the message myself, but find that Seaborne is busy jabbering away to Smith and blocking my path, I can cut through Wiese's office. If Wiese's door is locked, I'll take the back stairs down and back up, as I have seven different paths which I can take in my quest to arrive before the dean.

The mesh topology in the real world.

The mesh topology incorporates a network design in which each computer on the network connects to every other, creating a point-to-point connection between every device on the network. The purpose of such an implementation always centers on redundancy. If one cable fails, the data always has an alternative path to get to its destination. Each node can act as a relay. Because of the high cost in cabling such an implementation and issues with scalability, the mesh topology is not the first choice for most wired networks, but is much more popular with servers/routers.

Hybrid Topology
Whenever two topologies are mixed together, the hybrid label gets slapped on. This is often done when factors such as cost come into play. For example, you can use a mesh topology between servers, and then a star topology with workstations. While it would be ideal to use a mesh with the entire infrastructure, connecting every single workstation to every other is much too costly and the cost of a single failure is usually acceptable.