Ethernet and Link-Layer Switching
Ethernet is the dominant wired link-layer technology, defining a frame format, addressing, and a family of physical media, and it operates today as a full-duplex switched network rather than the shared, collision-prone medium of its origins.
Definition
Ethernet is a family of link-layer networking technologies standardized as IEEE 802.3, specifying frame format, MAC addressing, and physical-layer options, originally using CSMA/CD on a shared medium and now predominantly deployed as switched, full-duplex links.
Scope
This topic covers the Ethernet standard and its switched operation: the Ethernet frame format (preamble, source and destination MAC addresses, type, payload, and CRC), the evolution from CSMA/CD over shared coaxial cable to full-duplex switched twisted pair and fiber, the range of speeds from early megabit Ethernet to multi-gigabit variants, and the role of Ethernet switches in modern LANs and data centers. It complements the broader switched-LAN topic by focusing on Ethernet specifics, and excludes higher-layer routing.
Core questions
- What fields make up an Ethernet frame, and what does each do?
- How did Ethernet evolve from shared CSMA/CD media to switched full-duplex links?
- What range of speeds and physical media does the Ethernet family cover?
- How do Ethernet switches forward frames and eliminate collisions?
- Why has Ethernet remained dominant for wired local-area and data-center networks?
Key concepts
- Ethernet frame format
- MAC addresses
- CSMA/CD (legacy)
- full-duplex switched Ethernet
- physical media (twisted pair, fiber)
- Ethernet speeds (Fast, Gigabit, 10G+)
- CRC error detection
- Ethernet switches
Key theories
- Ethernet frame structure
- An Ethernet frame carries a preamble for synchronization, destination and source MAC addresses, a type/length field, a variable-length payload, and a CRC for error detection; this simple, stable format has persisted across decades of speed increases.
- From shared CSMA/CD to switched full duplex
- Original Ethernet shared a coaxial bus and used CSMA/CD to manage collisions; modern Ethernet gives each host a dedicated switched link operating in full duplex, eliminating collisions and letting capacity scale with switch port speeds.
Clinical relevance
Ethernet is the wired networking standard of offices, homes, and data centers, and its frame format is the substrate on which IP and the rest of the Internet stack ride over local links. Its switched, full-duplex form delivers the high, predictable throughput that data centers and enterprise networks depend on, and its continual scaling in speed keeps it central to modern infrastructure.
History
Metcalfe and Boggs invented Ethernet at Xerox PARC in 1973-1976, using CSMA/CD over a shared coaxial cable. It was standardized as IEEE 802.3 and progressed from 10 Mbps shared coax to twisted-pair wiring, then to switched full-duplex operation that removed collisions, and on through Fast, Gigabit, and multi-ten-gigabit speeds, outlasting competing LAN technologies such as Token Ring.
Key figures
- Robert Metcalfe
- David Boggs
Related topics
Seminal works
- metcalfe1976
- kurose2021
- ieee8023-2018
Frequently asked questions
- Does Ethernet still use CSMA/CD?
- Modern switched Ethernet generally does not. Each host has a dedicated full-duplex link to a switch, so there are no collisions and CSMA/CD is unnecessary. The mechanism remains part of the standard for legacy half-duplex shared segments, but those are rare today.
- Why has Ethernet outlasted other LAN technologies?
- Ethernet combined a simple, stable frame format with continual increases in speed, low cost, and easy switching. Its backward compatibility and ecosystem let it scale from megabits to many gigabits without changing the basic abstractions, allowing it to displace alternatives such as Token Ring and dominate wired networking.