Ethernet Fundamentals

What Ethernet is

Ethernet is the family of standards that defines how devices on a LAN send data to each other over cables. (Wi-Fi traffic also ends up on Ethernet, because access points bridge it onto the wired network.) Ethernet was developed in the 1970s, standardised by the IEEE as 802.3 in 1983, and has grown from 10 Mbps on shared coaxial cable to 400 Gbps and beyond on fibre. The frame format has barely changed in all that time, which is a big part of why Ethernet replaced its competitors.

If you have plugged a cable into a laptop, a games console or the back of a home router, you have used Ethernet. Every office, school and data centre is built on it.

💡 In simple terms: Ethernet is the set of rules for the “local roads” of a network. It decides what the vehicles (frames) look like, how they are addressed (MAC addresses), and what the roads are made of (cables and signals).

Why Ethernet exists

Before Ethernet, each computer maker had its own way of connecting machines, and equipment from different companies could not communicate. A LAN technology had to solve a few problems:

• A common language: any vendor's card must work with any vendor's switch. The IEEE 802.3 standard provides that.
• Addressing: on a shared network, each device must know which data is meant for it. MAC addresses solve this.
• Error detection: electrical noise can flip bits. Every frame carries a check value, so damage can be detected.
• Growth: the same frames had to work at 10 Mbps in 1983 and at 100 Gbps today.

Where Ethernet sits

Ethernet covers two layers of the OSI model:

Everything above Layer 2, such as IP, TCP and web traffic, travels inside Ethernet frames. Ethernet does not care what it carries; it just delivers frames across the local network.

The Ethernet LAN

An Ethernet LAN is a group of devices in one place (a home, an office floor, a building) connected by Ethernet. Today it almost always has a star shape: every device has its own cable to a switch.

Older Ethernet used a single shared coaxial cable, or a hub that repeated every signal to every port. Modern LANs use switches, which give each device its own full-speed, collision-free link.

The Ethernet frame

Every piece of data on an Ethernet link travels inside a frame like this:

• Destination and source MAC: which interface the frame is for, and which interface sent it.
• EtherType: what's inside (0x0800 IPv4, 0x86DD IPv6, 0x0806 ARP).
• Payload: usually an IP packet, 46–1500 bytes (the standard MTU).
• FCS (frame check sequence): a CRC checksum. If the receiver calculates a different value, it drops the frame.

A later lesson explains each field and decodes a real frame byte by byte.

Learn more: The Ethernet Frame

MAC addresses

Every Ethernet interface has a 48-bit MAC address. In a manufacturer-assigned address, the first half identifies the manufacturer (the OUI, Organizationally Unique Identifier), and the second half is unique to each interface:

Many phones and laptops now use random MAC addresses instead, for privacy. Address formats, special bits and random MAC addresses have their own lesson.

Learn more: MAC Addresses

How MAC-based delivery works, step by step

PC 1 (192.168.1.10, MAC 02:00:00:00:00:aa) wants to send a file to the server (192.168.1.20, MAC 02:00:00:00:00:dd) on the same LAN:

1. The application hands over the data. TCP and IP add their headers, and the IP packet is addressed to 192.168.1.20.
2. PC 1 finds the server's MAC address. The server is on the same subnet, so PC 1 checks its ARP cache, or asks with an ARP broadcast.
3. PC 1 builds a frame: destination 02:00:00:00:00:dd, source 02:00:00:00:00:aa, EtherType 0x0800, the IP packet, then the FCS.
4. The NIC sends the frame as signals over the cable to the switch.
5. The switch reads the destination MAC address, looks it up in its MAC address table and sends the frame out of the server's port only. It also learns from the source address that …:aa is on PC 1's port.
6. The server's NIC checks the frame: the destination is its own MAC address and the FCS is correct, so it passes the IP packet up to the operating system.

When the destination is on another network, the frame goes to the default gateway's MAC address instead, and the router builds a new frame for the next link. Two later lessons follow both journeys in full.

Learn more: Same-Subnet Communication · Different-Subnet Communication

Three types of destination

A later lesson explains these three types with real examples.

Learn more: Unicast, Broadcast and Multicast

A real-world example: a home network

In most homes, the “router” box from the internet provider contains a small Ethernet switch with four ports. A games console, a smart TV and a desktop PC plug into those ports. When the PC streams a film from a media server on the console, the frames go PC → built-in switch → console, delivered by MAC address: the internet is not involved at all. When the PC loads a website, its frames go to the router part of the box (the default gateway), which forwards the data to the internet.

Duplex and autonegotiation

Half duplex means a device can send or receive, but not both at once. It was used on old, shared Ethernet, where every hub formed one collision domain. Full duplex means sending and receiving at the same time, which is the norm on switched networks. Both ends normally use autonegotiation to agree on the best speed and duplex they both support.

⚠️ Duplex mismatch: if one end is hard-set to full duplex and the other autonegotiates, the autonegotiating side falls back to half duplex. The link comes up but is slow, with late collisions and CRC errors on the interface counters. Use auto on both ends, or set both ends to the same values.

A later lesson covers speed and duplex in detail, with the commands to check them.

Learn more: Speed and Duplex

When Ethernet fails

Useful commands

C:\>netsh interface show interface
Admin State    State          Type             Interface Name
-------------------------------------------------------------------------
Enabled        Connected      Dedicated        Ethernet
Enabled        Disconnected   Dedicated        Wi-Fi

$ ip -br link
lo               UNKNOWN        00:00:00:00:00:00 <LOOPBACK,UP,LOWER_UP>
enp3s0           UP             02:00:00:00:00:aa <BROADCAST,MULTICAST,UP,LOWER_UP>
wlp2s0           DOWN           02:00:00:00:00:bb <NO-CARRIER,BROADCAST,MULTICAST,UP>

Common mistakes

• Mixing up Ethernet and IP. Ethernet uses MAC addresses to deliver frames on the local network; IP uses IP addresses to deliver packets across networks. They work together.
• Thinking a switch understands IP. A plain (Layer 2) switch only reads MAC addresses. It forwards frames without looking at the IP packet inside.
• Assuming a link light means everything works. It only proves Layer 1. Speed, duplex, errors and IP settings still matter.
• Hard-setting speed or duplex on one end only. The other end falls back to half duplex, which causes a duplex mismatch.

Check yourself

Where to go next

Next, look inside the frame field by field, then study MAC addresses in detail.

Learn more: The Ethernet Frame · MAC Addresses