The Current Ethernet Specifications

Network Basics — How Bob Finds Sally

Two PCs, Bob and Sally, sit on the same LAN connected through a hub. How does Bob actually reach Sally? Not by name, and not by IP address alone — Bob ultimately needs Sally's MAC address, the hardware address burned into her network card.

Bob starts with name resolution (hostname-to-IP), normally handled by DNS. But if both hosts are on the same LAN, Bob can just broadcast the question directly — no DNS required. That's exactly what Windows NetBIOS name resolution does.

Packet capture, step by stepBob broadcasts a NetBIOS name query for "SALLY" to 192.168.0.255. Once the name resolves, Bob broadcasts an ARP request — "Who has 192.168.0.3?" — to the Ethernet broadcast address (all Fs). Sally replies directly to Bob with her MAC address. Only then can real communication begin.

Exam angleSally has to go through this exact same resolution process to talk back to Bob. Every host on a LAN independently discovers every other host's MAC address before Layer 2 delivery can happen — and we haven't even added a router yet.

Ethernet Basics

Ethernet is a contention media-access method — every host on a segment shares the same bandwidth and contends for the chance to transmit. It's popular because it's readily scalable (Fast Ethernet, Gigabit Ethernet slot into existing infrastructure), simple to implement, and straightforward to troubleshoot.

Ethernet spans two layers of the OSI model: the Data Link layer (addressing and framing) and the Physical layer (the actual signaling on the wire). This chapter covers both.

Collision Domain

A collision domain is a network segment where devices must pay attention to every transmission from every other device on that segment. If two devices transmit at the same instant, their signals interfere — a collision — and both must retransmit.

This scenario is classic in a hub environment: every port on a hub belongs to the same single collision domain (and the same single broadcast domain).

Broadcast Domain

A broadcast domain is the set of all devices that hear all broadcasts sent on a given segment. It's typically bounded by physical media — switches and repeaters — but can also describe a logical division where all hosts reach each other via a Data Link layer broadcast.

Exam angleOnly bridges, switches, and routers can prevent a transmission from propagating across an entire network. Hubs cannot — every port is one shared collision and broadcast domain.

CSMA/CD

Carrier Sense Multiple Access with Collision Detection (CSMA/CD) is Ethernet's contention-management protocol. A host checks the wire for a signal first; if it's clear, it transmits — and keeps monitoring the wire the whole time it's sending.

When a collision occurs

A jam signal informs all devices a collision occurred
A random backoff algorithm is invoked
Every device stops transmitting until its timer expires
All hosts have equal priority to transmit once timers expire

Effects of heavy collisions

Delay
Low throughput
Congestion

Exam angleBackoff is the enforced retransmission delay after a collision. If collisions keep happening after 15 attempts, the transmitting hosts time out.

Broadband vs. Baseband

Two ways to send signals down a wire.

Bit Rate vs. Baud Rate

Bit rate measures data bits (0s and 1s) transmitted per second — the modern, accurate unit (bps). Baud rate is an older term: one baud is one electronic state change per second, named after Jean-Maurice-Émile Baudot, who used it to measure telegraph speed.

Why bit rate replaced baudA single state change can carry more than one bit of data, so bps is the more accurate way to describe how much data is actually moving.

Wavelength & Duplex Modes

Wavelength is the distance between peaks in an electromagnetic wave — the property that lets us tell different kinds of energy (and different signaling channels) apart. Flip the cards below for the duplex facts the exam actually tests.

Exam angleHalf-duplex Ethernet (typically 10BaseT) shares a collision domain and runs only about 30–40% efficient in practice. Full-duplex uses a dedicated point-to-point link, has no collisions, and needs a dedicated switch port plus a NIC and switch port both capable of full-duplex.

Binary, Decimal & Hexadecimal Conversion

Ethernet MAC addresses — and later, IP addresses and subnet masks — are all rooted in binary. Master these conversions before Chapters 6–8.

Click bits on to build a byte and see its decimal value.

Decimal value: 0

Exam angleMemorize the subnet-mask byte values: 128, 192, 224, 240, 248, 252, 254, 255 (binary 10000000 through 11111111). Two hex characters always equal one byte — one nibble each.

Ethernet Addressing (MAC Addresses)

Every Ethernet NIC has a MAC (Media Access Control) address burned in — a 48-bit (6-byte) hardware address written in hex.

I/G + L/G bitsbit 47–46

Organizationally Unique Identifier (OUI)assigned by IEEE · 24 bits

Vendor Assignedunique per adapter · 24 bits

Individual/Group (I/G) bit

0 = individual MAC address (unicast). 1 = group address (multicast/broadcast).

Local/Global (L/G) bit

Tells you whether the MAC is the burned-in address (BIA) or one that's been changed locally — relevant later for IPv6 addressing.

Exam angleThe OUI is 24 bits (3 bytes), assigned by the IEEE to the manufacturer. The vendor-assigned low-order 24 bits start at all 0s for the first card made and count up to all 1s for the 16,777,216th card.

Ethernet Frames

The Data Link layer combines bits into bytes and bytes into frames, encapsulating Network-layer packets for transmission. Click each field for detail.

Exam angle802.3 uses a Length field; the Ethernet II frame uses a Type field to identify the upper-layer protocol (e.g., 0x0800 for IPv4, 0x86dd for IPv6). FCS provides error detection via CRC — not error correction.

Ethernet at the Physical Layer

Ethernet began with DIX (Digital, Intel, Xerox) at 10 Mbps over coax, then twisted-pair, then fiber. The IEEE 802.3 committee extended it through 802.3u (Fast Ethernet), 802.3ab (Gigabit over Cat 5+), and 802.3ae (10 Gbps). Browse the standards below.

Click a standard above for details.

Exam angle10Base2 (thinnet) and 10Base5 (thicknet) are pure legacy reference material — you'll never touch them in the field, but know their names, distances, and BNC/AUI connectors.

Common Ethernet Cable Types — Quick Reference

Table 4.4 condensed: speed, distance, and media for the standards that show up most on the exam.

Ethernet Name	Cable Type	Max Speed	Max Distance
10BaseT	UTP	10 Mbps	100 m/segment
100BaseTX	UTP, STP	100 Mbps	100 m/segment
10BaseFL	Fiber	10 Mbps	500–2,000 m
100BaseFX	MMF	100 Mbps	2,000 m
1000BaseT	UTP (4 pair, Cat 5+)	1000 Mbps	100 m
1000BaseTX	UTP (2 pair, Cat 6+)	1000 Mbps	100 m
1000BaseSX	MMF	1000 Mbps	up to 550 m
1000BaseCX	Shielded twinax	1000 Mbps	25 m
1000BaseLX	MMF/SMF	1000 Mbps	550 m MMF / 2,000 m SMF
10GBaseT	UTP	10 Gbps	100 m
10GBaseSR	MMF	10 Gbps	300 m
10GBaseLR	SMF	10 Gbps	10 km
10GBaseER	SMF	10 Gbps	40 km
40GBaseT	UTP Cat 8	40 Gbps	30 m

Real-world scenarioGigabit to the desktop plus VoIP, with 10 Gbps between access and core switches? Verify distances first. Under 100 m: Cat 5+ 1000BaseT to the desktop, 10GBaseT between switches. Beyond 100 m up to 400 m: 10GBaseSR over multimode fiber instead.

Ethernet over Other Standards

IEEE 1905.1-2013 defines a convergent home network spanning Wi-Fi, powerline, Ethernet, and coax. Two implementations worth knowing:

Exam angleBPL (IEEE 1901) can theoretically deliver 500+ Mbps over 1,500 meters of existing power wiring, but power lines are electrically noisy and can interfere with shortwave communications. HDMI Ethernet Channel piggybacks a data channel on the HDMI cable itself, up to 100 Mbps.

Exam Check

Eight quick questions pulled from this chapter's review section.

Score: 0 / 0