Ethernet

Ethernet is the dominant link-layer technology for wired local area networks (LANs).

It defines how frames are formatted and transmitted over the physical medium.

History

Ethernet was invented in the mid-1970s by Bob Metcalfe and David Boggs:

In the 1980s and early 1990s, Ethernet faced many challenges from other LAN technologies (Token Ring, FDDI, ATM).

Today, Ethernet is the dominant wired LAN technology.

Reasons for Ethernet's success:

By the late 1990s:

In the early 2000s:

Ethernet frame structure

Preamble Destination Address Source Address Type Data (Payload) CRC
8 bytes 6 bytes 6 bytes 2 bytes 46–1500 bytes 4 bytes
Field Description

Preamble

used for synchronization

  • starts with 10101010 repeated 7 times: lets the receiver align its clock to the sender's bit timing
  • ends with 10101011: marks the Start Frame Delimiter (SFD), indicating the start of the actual frame

needed because Ethernet has no shared clock between sender and receiver:

  • when adapter A sends bits (e.g., 10101010 10101010…), adapter B does not know:
    • where each bit starts
    • how long each bit lasts
  • even a tiny mismatch in clock speed will cause misalignment very quickly

Destination MAC address

identifies the receiving adapter (e.g., BB-BB-BB-BB-BB-BB)

Source MAC address

identifies the sending adapter (e.g., AA-AA-AA-AA-AA-AA)

Data

carries the IP datagram

MTU = 1500 bytes (larger datagrams must be fragmented at the IP layer)

minimum = 46 bytes:

  • smaller datagrams are padded ("stuffed")
  • IP uses its header length field to determine the actual datagram size and ignore padding

Type

indicates the upper-layer protocol (e.g. IP, ARP, Novell IPX, AppleTalk)

allows Ethernet to carry multiple ("to multiplex") network-layer protocols

serve to glue a protocol at one layer to a protocol at the layer above, analogous to:

  • the protocol field in the network-layer datagram
  • the port-number fields in the transport-layer segment

CRC

cyclic redundancy check (error-checking code)

Ethernet services

Ethernet provides a connectionless service to the network layer:

Ethernet provides an unreliable service to the network layer:

This lack of reliable transport at the link layer helps to make Ethernet simple and cheap.

Ethernet standards

Ethernet standards are defined by the IEEE 802.3 (Ethernet) working group and include many variants, such as:

Ethernet names follow a pattern:

Component Meaning Details

First number

Speed

  • 10 = 10 Mbps
  • 100 = 100 Mbps
  • 1000 = 1 Gbps
  • 10G = 10 Gbps

BASE

Baseband transmission

Baseband transmission means that the signal occupies the full bandwidth of the medium as a single channel (no frequency-division):

  • it mainly distinguishes from legacy broadband Ethernet (10BROAD36)
  • almost all IEEE 802.3 standards are for baseband Ethernet

Last part

Physical medium

Ethernet is both a link-layer and physical-layer technology and can run over different media:

  • T = twisted-pair copper cable
  • F / SX / LX / BX = fiber optics (varies by Ethernet generation)
  • earlier versions used coaxial cable (e.g., 10BASE-2)

Early Ethernet (like 10BASE-2 and 10BASE-5) used a shared coaxial bus. All devices shared the same medium, so collisions could happen. These were handled using CSMA/CD (collision detection).

Repeaters were used to extend signal distance by regenerating signals.

Modern Ethernet is different:

Despite major changes, Ethernet has stayed compatible. The core Ethernet frame format has remained compatible (with extensions such as VLAN tagging).

Previous Switched LANs and ARP All ⏎ Next Link-layer switches

A Kemar Joint