Want to keep learning?

This content is taken from the Raspberry Pi Foundation & National Centre for Computing Education's online course, An Introduction to Computer Networking for Teachers. Join the course to learn more.

Connection Models

The different arrangements of computer connections and devices are called topologies.

Point to point

In the simplest situation, one computer has a dedicated connection to another computer. Then the computer sending the information must simply choose the appropriate interface on which to send the data.

Before the advent of local area networks, or LANs in the home, two computers were often connected using RS232 cables, like this:

Two PCs directly connected by a single UTP cable.

To send data directly to a range of other devices, a dedicated connection to each one was necessary.

Mesh networks

A network where each device has a dedicated link to each other device is a mesh network. As the number of devices in such a network increases, so does the number of required connections to give each pair a dedicated link. Try to calculate how many links would be needed to connect three or more devices.

Number of devices Number of connections
2 1
3 3
4 6
5 10
n n(n-1)/2

n(n-1)/2: for n devices, there will be connections to (n-1) other devices. However, this counts every link twice, hence it’s necessary to divide by 2.

An animation showing a full mesh. Six computers are arranged at the corners of a regular hexagon. Each computer is directly connected to each other - two connections from each computer are sides of the hexagon, while the other 3 cross inside the hexagon. Signals travel along these connections between pairs of computers.

Mesh networks provide alternative paths for data in case of connection failure or connection overload. However, full meshed networks are costly, so partial meshes are used instead, in which only some of the possible connections are made.

Five devices, with some connections between them. A is connected to B and E, B is connected to A and D, C is connected to B and E, D is connected to B and E, and E is connecte to A, C and E.

Meshed networks are used extensively in some types of wireless network.

Ring topology

Ring topology connects each device to the next in a ring, around which the data travels. Each device has a unique address. The sender transmits the recipient address before the data. The recipient device then removes the data from the ring.

An animation of a ring network in action. Six computers are connected in a ring. These are labelled 1 to 6 clockwise around the ring. Computer 1 is highlighted with a box reading "Source: 1 Destination:5". A signal then travels clockwise around the ring, through computers 2, 3, and 4 until it reaches computer 5, where it stops.

Today, ring-based networks are rarely used, expect in some networks that span large geographic areas.

Bus topology and Ethernet

In the 1960s, a system of sending data over radio called Aloha was invented in Hawaii. All devices operated using the same frequency, so transmissions sent at the same time interfered and were lost.

The device that sent the original signal, on failing to receive an acknowledgement, would assume a collision had occurred. It would wait a random amount of time and then try sending again. This worked well for low volumes of traffic, but as demand increased, collisions occurred more frequently and the percentage of successful data transmissions fell.

Try the classroom activity in the PDF accompanying this step to illustrate the problem of signal interference and some possible solutions.

In the 1970s, an improved version of this network used a coaxial cable instead of airwaves. All of the computers were connected to this cable via a special interface cable, which plugged into a network interface controller (NIC) connected to the computer at one end, and tapped into the coaxial cable at the other end. This arrangement of devices is called a bus topology.

An image of the original Ethernet connection. An NIC at one station is shown, with the Controller connected to an Interface. The interface cable is then connected to a component labelled X, which is shown as a "tap" into a wire labelled "the ether". Other computers are shown with arrows pointing towards the wire. There are also arrows pointing each way along the wire. The wire is labelled with a "terminator" at each end.

This technology was called Ethernet, so named after the mythical ‘Ether’ that early scientists thought carried electromagnetic waves.

In subsequent versions of Ethernet, the network interface controller was put on a card inside the computer: the network interface card (also abbreviated as NIC).

The actual signal sent via the cable reaches all the connected devices. Like in a ring network, the signal contains a destination address to identify the recipient, and a source address to identify the sender.

An animation of a bus network in action. There is a horizontal "bus" wire. Six computers, labelled 1 to 6, are each connected to this wire by vertical wires. Computer 1 is highlighted with a box reading "Source: 1 Destination:5". A signal then travels from this computer to all of the other computers. Computer 5 is highlighted when it receives the signal.

This method of transmitting the signal via a wire also allowed the devices to detect if they were transmitting at the same time as another device. Devices could then stop and try again later. The ability to detect collisions meant that the cable was used much more effectively, and it led to the development of carrier-sense multiple access with collision detection (CSMA/CD), which we will look at in the next step.

Bus–based Ethernet networks were problematic because a failure anywhere on the bus brought the whole network to a stop. It was also difficult to locate the fault on the coaxial cable.

Star topology

In this topology, UTP cables connect each computer to a central device, which can send signals to all connected computers. In early star networks, the signal was sent to all destinations, meaning the networks behaved in a similar way to bus networks. However, with each computer having its own cable to the central device, it’s easier to find faults, because each link can be tested independently.

8 computers, each connected by a cable to a central device

Next up

In the next step, we will consider the addresses used to identify the devices on a network and the way this information is added to the data before it is sent via the wire. We will also look at how the device at the centre of the star network has evolved over time.

Questions

  • How many connections would you need to build a mesh network of ten computers?
  • Do you have experiences of the original bus-based Ethernet networks?
  • Why is Ethernet called a ‘best effort’ service?

Share your answers in the comments section below

Share this article:

This article is from the free online course:

An Introduction to Computer Networking for Teachers

Raspberry Pi Foundation