Ethernet operates at the link layer of TCP. It defines the physical media responsible for carrying data, the format of the data carried by that media and the hardware addressing between those devices. Therefore, it covers both the data link and physical layers of the OSI 7 layer model.

Ethernet defines the physical media used to carry data. The Ethernet standards are specified by the IEEE and belong to the 802.3 family of standards. They define physical properties of the cabling (for example whether it’s copper wires or a glass fibre) as well as data speeds.

Ethernet Standards

Some of the most common Ethernet varieties are summarised below. Mbps indicates a speed in megabits per second and Gbps indicates a speed in Gigabits per second. The term ‘BASE’ means that baseband signalling is used – the signal transmitted uses the full bandwidth of the media.

10BASE-T

Friendly Name: Ethernet

IEEE Standard: 802.3

Speed: 10 Mbps

Material: Copper

Maximum length: 100m

100BASE-TX (also referred to as 100BASE-T)

Friendly Name: Fast Ethernet

IEEE Standard: 802.3u

Speed: 100Mbps

Material: Copper

Maximum length: 100m

1000BASE-T

Friendly Name: Gigabit Ethernet

IEEE Standard: 802.3ab

Speed: 1 Gbps

Material: Copper

Maximum length: 100m

1000BASE-X

Friendly Name: Gigabit Ethernet

IEEE Standard: 802.3z

Speed: 1 Gbps

Material: Fibre

Maximum length: depends on fibre properties: 1000BASE-SX approx. 200m to 500m and 1000BASE-LX up to 5km

10GBASE-T

Friendly Name: 10 Gig Ethernet

IEEE Standard: 802.3an

Speed: 10Gbps

Material: Copper

Maximum length: 100m

How Ethernet Sends Data

Data is transmitted as an electrical signal with 2 different voltage levels representing the ‘1s’ and ‘0s’ of the data. Ethernet cables contain pairs of wires, with each pair forming a complete electrical circuit to carry the electrical signal.

Most modern Ethernet cables have 8 wires (in 4 pairs) and are terminated with RJ-45 connectors. This plug into RJ-45 sockets on your network devices - whether that is a NIC (Network Interface Card) in a server or a port on a network switch. We think of the wires in a cable as being numbered from 1 to 8.

All network devices and cables must comply with a standard way of using the pairs of wires inside an Ethernet cable.

Ethernet Pins

Historically, if we were connecting 2 devices (A and B), the Ethernet cable would use one pair of wires for transmitting from A to B (making B the receiver) and the other pair for transmitting from B to A (making A the receiver). In each pair, one wire would be positive and the other negative.

Therefore some device types (e.g. NICs) would expect to transmit on a certain pair (say 'Pair 1') and other device types (e.g. Switches) would expect to receive on a certain pair (say 'Pair 2'). This is fine if you want to connect a NIC to a Switch, but if you want to connect 2 Switches, they would get confused because they would both be trying to transmit on the same wires.

To solve this problem there are 2 different types of cable that you can get, a 'straight-through' for most applications, and a 'crossover' for where two of the same type of device (e.g. 2 Switches) are being connected.

You may see devices referred to as having 'medium dependant interfaces' (MDI) or 'medium dependent interface crossover' (MDI-X) interfaces.

This is most relevant for 10BASE-T and 100BASE-TX where devices transmit and receive as follows.

Devices that transmit on pins 1 and 2 (MDI):

  • Network Interface Cards (NICs)
  • Routers
  • Wireless Access Points

Transmit on pins 3 and 6 (MDI-X):

  • Hubs (rarely seen in modern networks)
  • Switches

Straight-through Ethernet Pin Out Example

 Device A            Device B
Pin 1  TX+   <-->   Pin 1  RX+
Pin 2  TX-   <-->   Pin 2  RX-
Pin 3  RX+   <-->   Pin 3  TX+
Pin 4  N/A   <-->   Pin 4  N/A
Pin 5  N/A   <-->   Pin 5  N/A
Pin 6  RX-   <-->   Pin 6  TX-
Pin 7  N/A   <-->   Pin 7  N/A
Pin 8  N/A   <-->   Pin 8  N/A

Here, Device A could be a PC NIC and Device B could be a Switch.

Typically used for the following connections:

  • PC/NIC to Switch
  • Wireless Access Point to Switch
  • Switch to Router

Crossover Ethernet Pin Out Example

 Device A            Device B
Pin 1  TX+   <-->   Pin 3  RX+
Pin 2  TX-   <-->   Pin 6  RX-
Pin 3  RX+   <-->   Pin 1  TX+
Pin 4  N/A   <-->   Pin 4  N/A
Pin 5  N/A   <-->   Pin 5  N/A
Pin 6  RX-   <-->   Pin 2  TX-
Pin 7  N/A   <-->   Pin 7  N/A
Pin 8  N/A   <-->   Pin 8  N/A

Here, Device A and Device B could both be PCs.

Typically used for the following connections:

  • PC to PC (NIC to NIC)
  • Switch to Switch

Modern Cabling

Modern cables (e.g. 1000BASE-T and upwards) use all pins for sending data and do not require one device to transmit and the other to receive on each pair. Furthermore, devices can usually automatically determine which pins to use which removes the need to worry about straight-through or crossover cables.

Unshielded Twisted Pair (UTP) Cable

UTP is a common and inexpensive choice for ethernet cabling. It is generally capable of transmitting data up to 100m.

Each pair of wires is twisted together so that any electromagnetic interference (EMI) is cancelled out. This interference could be ‘crosstalk’ from the other wires which make up the table.

Shielded Twisted Pair (STP) Cable

Twisted-pair cables can also be shielded to prevent EMI. Shielding can be applied to each pair of wires, to the whole cable (not the individual pairs) or to both the pairs and the whole cable.

Ethernet Standards vs Cable Categories

Ethernet standards (the 802.3 family) are specified by the IEEE and define the physical and datalink layers from the interface of one device to another. For example, on a 1000BASE-T link, the interface of the computer/device and switch as well as the cabling used in-between must adhere to the 1000BASE-T standard.

Ethernet cabling is typically sold under a category system defined by the Telecommunications Industry Association (TIA). The category defines the physical properties of the cable and will support multiple Ethernet standards (e.g. 10BASE-T, 100BASE-TX and 1000BASE-T). Cables belonging to these categories are commonly but not exclusively used for Ethernet data.

Cat 5 Ethernet Cable

Supported Ethernet Standards: 10BASE-T, 100BASE-TX, 1000BASE-T (although 5e is recommended for 1000BASE-T)

Maximum Speed: 1000 Mbps (under good conditions)

Maximum Bandwidth: 100 MHz

Cable type: UTP

Cat 5e Ethernet Cable

Offers improved mitigation of crosstalk compared to Cat 5 cable.

Supported Ethernet Standards: 10BASE-T, 100BASE-TX, 1000BASE-T

Maximum Speed: 1000 Mbps

Maximum Bandwidth: 100 MHz

Cable type: UTP

Cat 6 Ethernet Cable

Supported Ethernet Standards: 10BASE-T, 100BASE-TX, 1000BASE-T, 10GBASE-T

Maximum Speed: 10 Gbps

Maximum Bandwidth: 250 MHz

Cable type: UTP / STP

Cat 6a Ethernet Cable

Offers improved crosstalk and interference performance than Cat 6.

Supported Ethernet Standards: 10BASE-T, 100BASE-TX, 1000BASE-T, 10GBASE-T

Maximum Speed: 10 Gbps

Maximum Bandwidth: 500 MHz

Cable type: STP

Power over Ethernet (PoE)

Sometimes in a network, you may want to deliver power to devices using their Ethernet cable. This avoids the need for installing additional power sockets or running additional cables for power.

This is particularly useful for devices such as:

  • VOIP (Voice Over IP) Phones
  • Wireless Access Points
  • IP Cameras

Power over Ethernet (PoE) delivers DC power over 2 or 4 pairs of the wires in an Ethernet cable. Three types of PoE are standardized by IEEE 802.3: Alternative A, Alternative B, and 4PPoE.

In some implementations (using 10BASE-T or 100BASE-T), the power may be carried over the wires which are not used for transmitting data. In other implementations, wires will be used to carry power and data.

Fibre Optic Cables

Fibre optic cabling uses a fine fibre core made of glass or plastic to carry ‘1s’ and ‘0s’ as light. The network devices on either end will transmit the data using an LED or a laser and receive it using a photodetector. Fibre optic cables can generally support higher data speeds than copper.

The fibre optic core is surrounded by cladding to create the correct refractive properties for sending light along the fibre. To protect the fibre, a ‘buffer’ surrounds the cladding to prevent damage to the delicate materials. Finally, an additional plastic ‘jacket’ may be added.

There are two principal types of fibre optic cable: single-mode and multi-mode.

Single-Mode Fibre

  • Very small core diameter

  • Carries a single mode of light

  • More expensive

  • Can carry data over longer distances (e.g. 10s of kilometers)

Single-Mode Fibre Examples

  • 10GBASE-LR

  • 10GBASE-ER

  • 10GBASE-LX4

Multi-Mode Fibre

  • Larger core diameter

  • Carries multiple modes of light

  • Less expensive

  • Can only be used for shorter distances (e.g. 100s of meters)

Multi-Mode Fibre Examples

  • 10GBASE-SR

  • 10GBASE-LRM

  • 10GBASE-LX4

SFP Modules

In order to make network devices more flexible, enterprise devices may have SFP slots rather than, for example, RJ-45 sockets.

SFP stands for Small Form-factor Pluggable and it is possible to buy SFPs to support various types of Ethernet connector, including fibre and twisted pair.

SFP transceivers support speeds of up to 1 Gbit/s.

SFP+ transceivers support speeds of up to 10 Gbit/s.