Network topologies

Alright, now that you have been introduced to the internetworking devices, let's talk about topologies. First, let's define what a topology is. There are two types of topologies: you have the physical topology, which is how the network is physically connected. The other is the logical topology, which is how the path of the data flows. It depends on several factors, such as routing protocols, internetworking devices used, and the bandwidth configured on the interfaces of those internetworking devices.

But let's begin with the basics.

The Bus topology

Bus topologies use a primary cable, to which all end devices are connected. The data travels along this cable, hence the name Bus. The problem is that, at the time this type of topology existed, we were using coaxial cabling that at speeds of 10 Mbps, which is considered slow using today's standards. It was considered a shared medium, because the bandwidth was divided up based on how many computers you had connected. The following diagram shows the basic structure of Bus topology:

In this topology, Ethernet technology was used, which uses an access method called Carrier Sense Multiple Access Collision Detection (CSMA/CD). CSMA/CD is the method in Ethernet that end devices use, to be able to transmit their data. As I explained previously, if a device hears any noise on the wire, it will not transmit, it will wait until all noise is gone and then it will send its data. It could be that one node or device does not hear the other device, and both end devices are attempting to send at the same time. That will cause a collision; at that point, a jamming signal is sent, packets are dropped, and a countdown begins to see who transmits; the one whose countdown ends gets to send first.

So, imagine not the five nodes that you see in the figure, but hundreds of nodes trying to communicate. It's insane, since this type of topology creates only one collision domain and one broadcast domain that is running on half-duplex. It was not scalable and hard to troubleshoot, hence, not feasible at all.

Besides all that, if you do not terminate both ends of the cable, you will create something called reflection, which the signal that is on the wire reflect onto the cable continuously, creates noise so no one can transmit. The same thing would happen if your cable were cut somewhere in between; that is why troubleshooting this network was a nightmare. But, let's put the icing on the cake: if you don't ground one side of the cable, if a power surge hits your cable, it could fry all your nodes attached to the cable.

The Bus topology was not going to become the wave of the future.

The Star topology

In this topology, all devices are connected to a central device, in this case a layer-two switch. This is still using the Ethernet access method of CSMA/CD. But, since the media that is transferring the data is a switch, each port on a switch is a private collision domain, so you can have full-duplex, which will allow you to send and receive data. If one of the cables from an end device breaks, only that device will not be able to communicate on the network:

Even though you have increased the number of collision domains and they are private collision domains, which allows for greater bandwidth, one problem still exists: you have, by default, one broadcast domain. This means that when someone transmits on the network, everyone connected to that device, or to be more specific, VLAN 1, which is the native VLAN that all end devices connect to, will also hear that noise and still slow down your network.

The good news is that with a layer two or layer three switch, you can create multiple VLANs. You can logically segment your network so that when someone transmits within their own VLAN, no one else hears that noise.

To explain the obvious about this Star topology, you might be thinking, Hey, that doesn't look like a star, and you would be right. Just because they called it a Star, does not mean you are going to design your physical network in such a manner. It simply means you are connecting your devices to a central point where all devices can communicate:

The preceding illustration shows the reality of a common network design. You will run your cable from the office, cubicle, or classroom to the communications closet and terminate your cable at the patch panel. This in turn gets connected to the switch using patch cables, which then gets connected to the router.

With all that said and illustrated, I hope that clears up the Star topology definition.

The Ring topology

As illustrated in the following diagram, a token ring network is represented as a circle or ring, but there is more to token ring networks. A token ring network uses a central device called a Multi-Station Access Unit or Media Access Unit (MAU) and its purpose is to connect all end devices to it:

The MAU is not circular; it is rectangular and one could say it looks like a switch. There is a huge difference between them; an MAU has two ports called Ring in and Ring out to connect to other MAUs and concentrator ports for the end devices.

This MAU connects all these devices in a logical circular pattern, but the physical topology is that of a star.

The type of access method is called token passing and is deterministic in nature, unlike Ethernet which is contention-based. By this, I mean a token ring has an empty, free-flowing token that goes around the network waiting for someone to seize the token and send data. Only the person with that token can transmit, and once the token is seized, no other token is generated. Therefore, no one else can transmit until that token has been released by the destination end device back into the network.

With the token ring, there are no collisions and it was reliable, but the speed of it was just too slow. Again, the popularity for designing, implementing, and using a token ring network simply did not catch on for use on LANs.

On WANs, we did have the Fiber Distributed Data Interface (FDDI) which used token ring technology and ran it up to gigabit speed. But, as you go through this book, the token ring will not be mentioned at all; it is considered an older technology and, for LANs, it is not used. Also, for your certification you will not need to know this information. Just think of it as information to have in your back pocket for interviews and dinner parties.