Networking: Ethernet

Ethernet is the name given to a set of standards used for designing networks. Ethernet is the most common network type used for connecting computers and other devices. It was was invented in 1972 by future 3Com founder Bob Metcalfe at the Xerox Palo Alto Research Center (Xerox PARC).

An Ethernet network can be compared to a huge number of conversations between an even huger number of people. There are, however, a few rules governing this conversation:

• Only one person can speak at a time. If two people start to speak simultaneously, both of them stop talking, wait a moment and then try to talk again
• When not being directly addressed, all participants should keep their hands over their ears so they can’t hear the conversation
• When someone has something to say, the speaker points at the intended listener, who uncovers their ears to hear. When the speaker is done, the listener must cover their ears again
• Sometimes a speaker will have something to say to everyone simultaneously. In that case, the speaker points at everyone, and everyone uncovers their ears to hear the message

This is technically known as Carrier Sense, Multiple Access, Collision Detection. What it means is that NICs are constantly listening to the network and are able to tell if a piece of data has collided with another from elsewhere. When this happens, the data is lost and both transmitting NICs will wait for a random amount of time (in milliseconds) and retransmit the data, where it will hopefully arrive intact this time around.

Ethernet is fine for low to medium levels of network activity, but when an increasing number of conversations creates a lot of traffic, ethernet will get bogged down due to increasing collisions and retransmissions. The user's experience of this is a slow network.

Most of the time Ethernet enables orderly conversations between computers on a network. Sometimes one of those computers will broadcast a message to every other computer on the network. These broadcasts are usually related to network administration, such as routers advertising their presence or computers requesting IP addresses.

Every network interface card (NIC) has a unique embedded hardware address called a MAC (Media Access Control) address, which makes it individually addressable. When one computer wants to talk to another, it puts a message on the network aimed at the intended recipient’s MAC address. The network card of the intended receiver hears the message, accepts the data and passes it to the computer for processing by some application. The rest of the network cards ignore the message.

5,4,3,2,1

Ethernet has a rule known as the 5:4:3 rule, for the number of repeaters and segments on shared access Ethernet backbones in a tree topology. The 5:4:3 rule divides the network into two types of physical segments: populated (user) segments, and unpopulated (link) segments. User segments have users' systems connected to them. Link segments are used to connect the network's repeaters together. The rule dictates that between any two nodes on the network, there can only be a maximum of five segments, connected through four repeaters, or concentrators, and only three of the five segments may contain user connections.

The Ethernet protocol requires that a signal sent out over the LAN reach every part of the network within a specified length of time. The 5:4:3 rule ensures this. Each repeater that a signal goes through adds a small amount of time to the process, so the rule is designed to minimise transmission times of the signals.

The 5:4:3 rule only applies to shared-access Ethernet backbones. A switched Ethernet network should be exempt from the 5:4:3 rule because each switch has a buffer to temporarily store data and all nodes can access a switched Ethernet LAN simultaneously.