Ethernet is a family of standards for moving data across local wired networks. A device places network data inside an Ethernet frame, identifies source and destination interfaces with MAC addresses, and sends an encoded signal through copper or optical media. An Ethernet switch reads the destination address and forwards the frame toward the correct port. Ethernet carries local traffic; it is not itself the entire internet.
What happens in five steps
1. A device builds an Ethernet frame
When the network stack has data for a local link, it wraps that data in a frame. The frame includes a destination MAC address, source MAC address, a field identifying the carried protocol, the payload, and a frame check sequence. Preamble and delimiter bits help the receiver synchronize with the beginning. The exact handling can include virtual-LAN tags or other options, but the basic job is to create a bounded local-link message.
2. The network interface encodes a physical signal
The Ethernet controller passes the frame toward a physical-layer transceiver. On twisted-pair copper, electronics convert the bits into carefully timed electrical symbols sent across wire pairs. Fiber Ethernet uses modulated light. Different Ethernet variants define different speeds, cable categories, wavelengths, lane counts, and maximum link lengths. Both ends must support a compatible mode and negotiate or be configured correctly.
3. A switch receives and examines the frame
A switch verifies enough of the incoming signal to recover a frame and reads its source and destination MAC addresses. By observing source addresses, it learns which devices are reachable through which ports. That learned table is temporary and changes as devices move or entries age. The switch does not normally inspect a web page’s meaning to perform basic forwarding; it makes a local link-layer decision.
4. The switch selects an outgoing port
If the destination address is known, the switch sends the frame only through the associated port, unless policy says otherwise. Broadcast traffic and unknown destinations may be flooded to the relevant local segment. Each switched link is a separate full-duplex path in modern networks, so devices can transmit and receive without the shared-wire collisions associated with early Ethernet. A loop-control design is still important because duplicated paths can circulate frames.
5. The receiver checks and unwraps the frame
The receiving interface calculates the frame check value and compares it with the transmitted value. A damaged frame is discarded rather than silently delivered as good data. If the destination and protocol are appropriate, the payload moves up the network stack. Recovery from a lost frame generally belongs to higher-layer protocols or applications; ordinary Ethernet does not guarantee that every frame will arrive.
What a MAC address does
A MAC address identifies an interface for delivery on an Ethernet link. It is different from an IP address, which supports routing across networks. Local systems use mechanisms such as ARP for IPv4 or Neighbor Discovery for IPv6 to determine which link-layer destination corresponds to a nearby IP destination or default router. Addresses can be changed, virtualized, randomized in some contexts, or shared by special systems, so a printed address is not a permanent identity for a person.
Switching and routing are different stages
An Ethernet switch forwards local frames using MAC information. A router forwards IP packets between networks using IP destinations. When a computer sends traffic to a remote website, its Ethernet frame is normally addressed to the local router, while the IP packet inside names the remote destination. The router removes the incoming local frame, makes a routing decision, and creates suitable framing for the next link.
Why Ethernet is not the same as internet service
An Ethernet cable can provide a perfectly healthy link to a router even if the modem, ONT, provider account, DNS service, or upstream route has failed. Link lights show electrical or optical connectivity and often activity, not universal internet reachability. This is the wired counterpart to a phone reporting connected without internet over Wi-Fi.
Why cable category and link speed matter
Twisted-pair Ethernet depends on controlled cable construction and termination. Category ratings describe performance requirements, not a magical speed embedded in the jacket. Link length, connector quality, pair integrity, electromagnetic noise, and the capabilities of both ports affect the negotiated rate and reliability. A link may fall back to a lower speed when one pair is damaged or hardware lacks a faster mode. The applicable equipment manuals and cabling standards govern installation limits.
What full duplex means
Modern switched Ethernet commonly dedicates a link between one device and one switch port and supports simultaneous traffic in both directions. That is full duplex. Early shared coaxial networks and hubs placed devices in one collision domain, where carrier sensing and collision detection controlled access. The historical name remains in the standard family, but collisions are not the normal explanation for performance on correctly configured full-duplex switched links.
Ethernet compared with Wi-Fi
Wi-Fi and Ethernet both provide local network links, but their physical access methods differ. Ethernet usually uses a dedicated cable path, while Wi-Fi devices share radio airtime and must contend with interference, distance, walls, and neighboring networks. Wi-Fi frames are not simply radio versions of every Ethernet frame, though access points bridge traffic between the two systems. Wired Ethernet often offers stable latency and capacity; Wi-Fi offers mobility and easier placement.
How Power over Ethernet fits in
Power over Ethernet can deliver electrical power and data over compatible cabling to devices such as access points, cameras, or phones. The power-sourcing equipment and powered device detect and negotiate within defined rules before normal power is applied. Compatibility, cable quality, heat, and power class matter. Do not assume that an unmarked passive injector follows standardized detection or is safe for arbitrary equipment.
Installation and safety boundaries
Use cable and connectors rated for the environment, building code, and required fire classification. Keep ordinary network cabling separated from hazardous conductors as required, and do not route it through spaces or outdoors unless the cable and protection are suitable. Building-entry, lightning, industrial, and in-wall work can require qualified installers. Never handle damaged wiring during a storm, and never connect Ethernet equipment to telephone, mains-power, or unidentified jacks merely because a plug appears to fit.
The useful mental model
Think of Ethernet as addressed envelopes on a local wired system. The network interface creates and signals each envelope, a switch learns which door leads to each local address, and the receiver checks whether the envelope arrived intact. A router takes the payload beyond that local neighborhood. Cable speed, switching, routing, and internet access are connected parts, but each answers a different question.
