6.1. Ethernet Basics

Origins of Ethernet

Ethernet is the most common technology used to build local area networks in homes, offices, and data centers. It defines how devices in the same local network send and receive data. When a laptop connects to a switch with a cable, or a desktop connects to a wall port in an office, it is almost always using Ethernet.

Ethernet began as a way to connect computers over a shared cable. Over time it evolved into a family of standards that describe different speeds, cable types, and frame formats. Today, when people say “wired networking” they usually mean Ethernet, even though many versions exist under that single name.

Ethernet is defined by the IEEE 802.3 standard. This standard describes the rules for how devices access the medium, how bits are encoded on cables, how frames are structured, and what speeds and physical media are supported. Later chapters will go deeper into MAC addresses and Ethernet frames, so here we focus on what makes Ethernet itself special and how it behaves on a basic level.

What Ethernet Provides

Ethernet operates at the data link and physical layers in the OSI model. It provides a way for devices on the same local segment to deliver frames to each other using hardware addresses. Higher layer protocols such as IP and TCP rely on Ethernet to move their packets across the local network.

At a high level, Ethernet gives you three things on a local network segment. It gives you addressing using MAC addresses. It gives you encapsulation of data from upper layers into Ethernet frames. It gives you rules for how the medium is used so that devices do not talk over each other in an uncontrolled way.

It is important to understand that Ethernet itself does not route between networks. It only delivers frames within a single broadcast domain. Routing and IP addressing are handled at higher layers and by routers, which you will study in separate chapters. Here the focus is on how Ethernet works inside one local network.

Shared Media and Early Ethernet

The earliest Ethernet networks used a single shared cable that many devices tapped into. This is often called “shared media” Ethernet. All devices were connected to the same physical medium, so if two devices tried to send data at the same time their signals would collide.

To manage this, classic Ethernet used a method called Carrier Sense Multiple Access with Collision Detection, or CSMA/CD. Devices listened to the cable before transmitting. If they detected another transmission they waited. If two devices still transmitted at the same time and a collision occurred, they both backed off for a random period then tried again.

CSMA/CD allowed multiple devices to share one cable without a central controller. However, it limited performance. Only one device could successfully send at a time, and collisions wasted bandwidth. Network size and cable length were also limited because collision detection needed to work reliably over the entire shared medium.

Hubs, which appeared later, kept the idea of a shared medium. All ports on a hub shared the same collision domain, so CSMA/CD still applied. Modern networks have almost completely replaced hubs with switches, which changes how Ethernet behaves and removes collisions in most cases.

Half Duplex and CSMA/CD

In a half duplex Ethernet environment, a device can either send or receive at one time, but not both at once. This matched the idea of a shared medium, since a device needed to listen for other traffic and detect collisions.

In such a network, the rules look like this. A device first listens to see if the line is idle. If the line is quiet, it starts transmitting its frame. While transmitting, it still monitors the line to see if what it sends matches what it hears. If it detects a mismatch, that means a collision occurred. Both transmitters stop, send a brief jamming signal, then wait a random time before trying again.

The use of carrier sense, multiple access, and collision detection is summarized by the CSMA/CD name. This mechanism is specific to shared, half duplex Ethernet. Its purpose is to coordinate access to the medium so that multiple devices can share it without constant interference.

With the transition to switches and full duplex links, CSMA/CD became mostly unnecessary. Each link between a device and a switch is its own collision domain. Two devices are no longer trying to talk over the same cable at the same time. For this reason, modern Ethernet networks with switches typically operate in full duplex mode and do not use CSMA/CD in practice, even though the concept still belongs to the Ethernet standard.

Full Duplex and Modern Ethernet Links

Modern Ethernet links between a device and a switch port are almost always full duplex. This means a device can send and receive at the same time over separate paths within the cable. In full duplex mode, collisions do not occur because the sending and receiving signals are kept logically separate on different wire pairs or fiber strands.

When an Ethernet interface comes up, it can negotiate speed and duplex with the device on the other end. This is called autonegotiation. For example, a network card and a switch port might negotiate to use 1 Gbit/s full duplex. If both sides support the same capabilities, they agree on the highest common speed and on full duplex whenever possible.

Because modern Ethernet uses full duplex and point to point links, the network behaves more predictably than early shared Ethernet. There is no need to wait for an idle line, and there are no collisions, so CSMA/CD backoff logic is not used. This helps to increase efficiency and reduces delays.

It is still possible to misconfigure duplex on older equipment. If one side believes it is full duplex and the other believes it is half duplex, performance problems and errors can appear. While such mismatches are less common today, understanding the difference between half and full duplex helps in troubleshooting older or mixed environments.

Ethernet Speeds and Naming

Ethernet includes many standardized speeds. Each speed level has a common shorthand name and a more formal standard name. Different speeds also have different media options, such as copper or fiber. The following table gives a simple overview of some common Ethernet speeds and typical names.

The naming has a pattern. The number at the front shows the speed in Mbit/s or Gbit/s. The word BASE indicates baseband signaling, which means the medium carries only Ethernet signals, not multiple signals at once. The last part such as T, TX, SX, or SR describes the physical medium and encoding type, for example twisted pair copper or short range fiber.

Over time, Ethernet has grown beyond 10 Gbit/s. There are standards for 25, 40, 50, 100 Gbit/s and beyond, especially in data centers. The same basic Ethernet frame format remains in use, but the physical layer and encoding schemes change so that more bits can be pushed through the medium each second.

Although the physical details change, all these versions are still considered Ethernet. This backward compatibility at higher layers is one of the main reasons Ethernet remained dominant over other local networking technologies.

Frames and Maximum Transmission Unit

Ethernet sends data in units called frames. A frame includes headers, payload, and a trailer, but the exact structure is explained in detail in a later chapter. Here the key point is the typical maximum payload size that Ethernet allows, which affects how upper layer protocols divide their data.

The normal maximum payload for a standard Ethernet frame is 1500 bytes. This is part of what is called the Maximum Transmission Unit, or MTU, for Ethernet. The MTU tells you the largest amount of upper layer data that can be carried inside one Ethernet frame without fragmentation at that layer.

The MTU value is important for protocols like IP, which must know how much data can fit inside a single Ethernet frame. If a higher layer packet is larger than the MTU of the underlying Ethernet network, it must be split into smaller pieces or fragmentation will occur at some point in its path.

Networks can support different MTU values if they are configured for special frame types such as jumbo frames. Jumbo frames allow payloads larger than 1500 bytes, often up to around 9000 bytes. These are used in specific environments such as data centers or storage networks. They are not universal and require end to end support.

For beginners, it is enough to remember that a typical Ethernet network uses an MTU of 1500 bytes, and that this size influences how data is chunked and transmitted by higher layer protocols across the network.

Broadcast and Unicast Behavior in Ethernet

Ethernet supports different kinds of frame delivery across a local network segment. The most basic types are unicast, broadcast, and multicast. These categories depend on the destination MAC address used in the frame.

Unicast frames have a destination MAC address that identifies a single network interface. In a switched Ethernet network, switches learn which MAC addresses are reachable on which ports. Frames with a unicast destination are forwarded only out the port that leads to the correct device. This makes unicast traffic efficient because it does not reach every device.

Broadcast frames use a special MAC address of all ones in binary. In hexadecimal notation this is written as FF:FF:FF:FF:FF:FF. When a device sends a frame to this broadcast MAC address, every device in the same broadcast domain receives and processes it. Broadcasts are needed for some basic functions, such as discovering the MAC address that corresponds to an IP address, which is handled by ARP in IPv4 networks.

Multicast frames use MAC addresses reserved for groups of devices rather than for a single interface or for everyone. If a device joins a particular multicast group, it processes frames sent to that group MAC. Other devices that have not joined the group can ignore this traffic.

The important concept at this stage is that Ethernet uses MAC addresses to control which devices should look at a frame. A unicast address targets one device, a broadcast address targets all devices in the broadcast domain, and a multicast address targets a selected group. Switches use these addresses to decide how to forward or flood traffic across the local network.

Relationship of Ethernet to Higher Layers

Ethernet does not know anything about IP addresses, ports, or applications. Its job is to deliver frames between directly connected devices in a local network. Higher layer protocols sit inside the payload of Ethernet frames and rely on Ethernet for delivery.

When an IP packet is sent on a local Ethernet network, it is placed inside an Ethernet frame. The Ethernet header contains the source and destination MAC addresses and an indication that the payload is an IP packet. The receiving device extracts the IP packet and passes it up the stack without Ethernet itself interpreting IP fields.

Because Ethernet is ignorant of higher layer details, the same Ethernet network can carry many kinds of traffic at once. It can carry IPv4, IPv6, ARP, and other protocols simultaneously, as long as each protocol has a defined way to identify itself in the Ethernet header. This flexibility helped Ethernet outlast other local networking technologies.

VLANs, trunking, and other advanced Ethernet based features extend the basic Ethernet concept without changing its fundamental role. They allow larger or more complex networks to be built while still using Ethernet at the data link and physical layers. These topics are covered later. For Ethernet basics, it is enough to understand that Ethernet is the common foundation that upper layer protocols use to move data within a local network.

Summary of Key Ethernet Ideas

Ethernet is the dominant technology for local wired networking, defined by the IEEE 802.3 standard. It started as a shared medium using CSMA/CD in half duplex mode and evolved into point to point, full duplex links that operate without collisions. It supports a wide range of speeds and media types, from 10 Mbit/s copper links to multi gigabit fiber connections.

Ethernet transports data in frames, with a typical MTU of 1500 bytes for the payload that carries upper layer information. It uses MAC addresses to deliver frames as unicast, broadcast, or multicast within a broadcast domain. Higher layer protocols such as IP and TCP depend on Ethernet for their local delivery but are independent of its physical details.

Later chapters will examine MAC addresses, Ethernet frame structure, collision and broadcast domains, and the way switches control traffic in much more detail. The concepts in this chapter provide the basic mental model that Ethernet is a data link technology used to move frames between neighboring devices within a local network.

Views: 80