3 Ethernet LANs, Cabling and Interfaces · CCNA

Exam topics covered

1.2 Describe characteristics of network topology architectures (star, mesh topologies introduced)

1.3 Compare physical interface and cabling types

1.3.a Single-mode fiber, multimode fiber, copper · 1.3.b Connections (Ethernet shared media and point-to-point)

1.4 Identify interface and cable issues (collisions, errors, mismatch duplex, and/or speed)

1.1.h PoE (verification)

Foundation. Everything in a network ultimately travels over a physical link. If the cable is the wrong type, too long, damaged or misconfigured at either end, nothing above it works—and the symptoms (slow file transfers, dropped calls, intermittent pings) often look like higher-layer problems. This chapter covers the Ethernet family of standards, copper and fibre cabling, connectors and transceivers, MAC addressing, and the speed and duplex settings that must agree on both ends of a link. You will learn to read the counters in show interfaces and to diagnose the classic duplex mismatch, a favourite exam scenario.

3.1 Ethernet: Shared Media and Point-to-Point

Ethernet was invented in the 1970s as a shared medium: every station was attached to a single coaxial cable, and any frame sent was heard by all. Because only one station could transmit at a time, Ethernet used CSMA/CD (Carrier Sense Multiple Access with Collision Detection): listen before sending; if two stations transmit at once, both detect the collision, send a jam signal, wait a random backoff time and retry. Hubs preserved this behaviour with twisted-pair cabling: a hub repeats every bit out every other port, so all attached devices form one collision domain and must use half duplex.

Modern Ethernet is built from point-to-point links: each cable connects exactly two devices, typically an endpoint and a switch port, or two switches. A point-to-point link can run full duplex, sending and receiving simultaneously on separate wire pairs or fibres, so collisions are impossible and CSMA/CD is disabled. A 1 Gbps full-duplex link can carry 1 Gbps in each direction at once.

Shared media (hub) Point-to-point (switch)

one collision domain, half duplex, CSMA/CD collision domain per port, full duplex every frame repeated to every port frames forwarded only where needed

Figure 3-1 Shared media versus point-to-point Ethernet.

3.1.1 Ethernet standards

Ethernet standards are defined by the IEEE 802.3 working group. Each has an informal name such as 1000BASE-T: the number is the speed in Mbps (or G for Gbps), BASE means baseband signalling, and the suffix identifies the medium (T = twisted pair; S/L/E = short/long/extended wavelength fibre; numbers such as 4 indicate parallel lanes). Table 3-1 lists the ones you are most likely to meet.

Table 3-1 Common Ethernet standards

Name IEEE Speed Medium Max distance (typical)
10BASE-T 802.3i 10 Mbps Cat3+ UTP, 2 pairs 100 m
100BASE-TX (Fast
Ethernet)
802.3u 100 Mbps Cat5+ UTP, 2 pairs 100 m
1000BASE-T (Gigabit) 802.3ab 1 Gbps Cat5e+ UTP, 4 pairs 100 m
1000BASE-SX / LX 802.3z 1 Gbps MMF / SMF (LX also
MMF)
~550 m / 5 km (SMF)

2.5gbase-t /

5gbase-t

10GBASE-SR / LR 802.3ae 10 Gbps MMF / SMF 300–400 m (OM3/OM4) / 10 km

25GBASE-SR 802.3by 25 Gbps MMF ~100 m (OM4)

40GBASE-SR4 / LR4 802.3ba 40 Gbps MMF (4 lanes) /

Smf

~150 m / 10 km
100GBASE-SR4 / LR4 802.3bm /
ba
100 Gbps MMF (4 lanes) /

Smf

~100 m / 10 km

All BASE-T twisted-pair standards share the same 100 metre maximum segment length (90 m permanent link plus 10 m of patch cords). If a link must be longer, use fibre. 10BASE-T and 100BASE-TX use two pairs; 1000BASE-T and faster use all four pairs.

3.2 Copper Cabling

Ethernet copper cabling is unshielded twisted pair (UTP): four pairs of insulated copper wires, each pair twisted together to cancel electromagnetic interference (EMI) and crosstalk from neighbouring pairs. Shielded variants (STP, F/UTP, S/FTP) add foil or braid for electrically noisy environments. Cable quality is graded by category, which defines the frequency the cable supports and therefore the speed it can carry over 100 m.

Table 3-2 UTP categories

Category Bandwidth Typical Ethernet use
Cat5 100 MHz 100BASE-TX (obsolete for new installs)
Cat5e 100 MHz 1000BASE-T; 2.5GBASE-T
Cat6 250 MHz 1000BASE-T, 5GBASE-T; 10GBASE-T up to ~55 m
Cat6a 500 MHz 10GBASE-T to 100 m; common for new AP cabling
Cat8 2000 MHz 25/40GBASE-T to 30 m (data centre)
3.2.1 RJ-45 pinouts: straight-through and crossover

UTP is terminated with 8-pin RJ-45 connectors (formally 8P8C) wired to one of two TIA/EIA-568 colour schemes, T568A or T568B. They differ only by swapping the green and orange pairs. For 10BASE-T and 100BASE-TX, only pins 1, 2, 3 and 6 are used. PCs, routers and wireless APs transmit on pins 1,2 and receive on 3,6. Switches and hubs do the opposite: transmit on 3,6 and receive on 1,2.

A straight-through cable uses the same standard on both ends, so pin 1 connects to pin

1. It joins unlike devices: PC to switch, router to switch.

A crossover cable uses T568A on one end and T568B on the other, so pins 1,2 connect to 3,6. It joins like devices: switch to switch, PC to PC, router to router, and PC to router. Straight-through (PC to switch) Crossover (switch to switch)

PC NIC Switch port

Tx+ → Rx+
Rx+ ← Tx+
Tx Rx (3,6)

Dashed: pins 4, 5, 7, 8 unused at 10/100 Mbps (used by 1000BASE-T)

Figure 3-2 10/100 Mbps pinouts. A straight-through cable connects pin n to pin n; a crossover connects 1↔3 and 2↔6 so that

each side's transmit pair reaches the other's receive pair.

Table 3-3 T568A and T568B colour codes

Pin 1 2 3 4 5 6 7 8

T568A White/green Green White/orange Blue White/blue Orange White/brown Brown T568B White/orange Orange White/green Blue White/blue Green White/brown Brown Gigabit Ethernet and faster use all four pairs in both directions at once, so a gigabit crossover also swaps pins 4,5 with 7,8. In practice you will rarely need a crossover cable today, thanks to Auto-MDIX (automatic medium-dependent interface crossover): the port detects which pairs the other side transmits on and swaps its own pairs internally. Auto-MDIX is enabled by default on modern Catalyst switches, but it relies on speed autonegotiation; if you hard-code speed and duplex on some platforms, Auto-MDIX may stop working and a wrong cable will leave the link down.

Exam alert

Exam questions still ask which cable to use: straight-through for unlike devices (host or router to switch), crossover for like devices (switch to switch, host to host, router to router, and router to host). A router and a PC transmit on the same pins, so they count as “like” devices.

3.3 Fibre-Optic Cabling

Fibre carries data as pulses of light in a glass core surrounded by cladding with a lower refractive index, which keeps the light inside by total internal reflection. Fibre is immune to EMI, cannot be tapped easily, does not carry electricity between buildings (avoiding ground-loop and lightning problems) and supports far longer distances and higher speeds than copper. Each link normally uses a pair of fibres, one per direction.

Figure 3-3 Single-mode and multimode fibre cross-sections (not to scale).

Multimode fibre (MMF) has a wider core, so light travels along many paths (modes) that arrive at slightly different times. This modal dispersion limits distance to a few hundred metres, but the wide core allows cheaper LED or VCSEL light sources. MMF grades are labelled OM1 to OM5; OM3 and OM4 (aqua jacket) are common for 10 Gbps and above. Single-mode fibre (SMF) has a tiny core that carries a single mode, driven by a more precise and more expensive laser, and reaches 10 km, 40 km or more. SMF jackets are usually yellow.

Table 3-4 Copper versus fibre

The most common fibre connectors are LC (Lucent Connector), a small push-pull connector usually used in duplex pairs and standard on SFP modules, and SC (Subscriber/Standard Connector), a larger square push-pull connector common on older equipment and patch panels. You may also see ST (bayonet) and MPO/MTP (multi-fibre, for 40/100G parallel optics). Switches rarely have fibre optics built in. Instead they have transceiver slots that accept hot-swappable modules: SFP (1 Gbps), SFP+ (10 Gbps), SFP28 (25 Gbps), QSFP+ (40 Gbps) and QSFP28 (100 Gbps). You choose the module to match the fibre type and distance, for example a GLC-SX-MMD (1000BASE-SX) for multimode or an SFP-10G-LR for single-mode. Copper SFPs (1000BASE-T) and direct-attach cables (DAC) also exist for short runs.

Jacket colours: aqua = OM3/OM4 multimode, yellow = single-mode, orange = older OM1/OM2

Figure 3-4 An SFP transceiver with an LC duplex receptacle, and LC and SC connectors.

Watch out

Both ends of a fibre link must use the same fibre type and compatible optics (SX to SX, LR to LR), and the transmit fibre at one end must reach the receive port at the other. A link that stays down with fibre often means swapped Tx/Rx strands, mismatched optics, or a dirty connector.

3.4 Ethernet MAC Addresses

Every Ethernet interface has a 48-bit MAC address, also called a burned-in address (BIA) because it is assigned by the manufacturer. It is written as 12 hexadecimal digits: Cisco uses 0050.7966.6800 , Windows 00-50-79-66-68-00 , and Linux/macOS 00:50:79:66:68:00 . The first 24 bits are the Organizationally Unique Identifier (OUI), assigned to the vendor by the IEEE; the vendor assigns the last 24 bits uniquely to each interface.

00 : 50 : 79 66 : 68 : 00
OUI (24 bits) — vendor Vendor-assigned (24 bits)
First byte 0000 0000
last bit (I/G): 0 = unicast, 1 = group (multicast/broadcast)

second-last bit (U/L): 0 = globally unique, 1 = locally administered

Figure 3-5 MAC address structure. The two low-order bits of the first byte identify group and locally administered addresses. Ethernet destination addresses come in three types:

Unicast — identifies a single interface. Source MAC addresses are always unicast. Broadcast — FFFF.FFFF.FFFF , delivered to every device in the broadcast domain (used by ARP requests and DHCP Discover).

Multicast — identifies a group of interested devices. The I/G bit is 1. IPv4 multicasts map to 0100.5Exx.xxxx , IPv6 multicasts to 3333.xxxx.xxxx , and protocols such as STP use reserved addresses like 0180.C200.0000 .

3.5 Speed, Duplex and Autonegotiation

Both ends of a link must agree on speed and duplex. Autonegotiation (IEEE 802.3u, extended in 802.3ab) lets the two ends exchange their capabilities using fast link pulses and choose the best common mode, preferring higher speed and then full duplex. Cisco switch ports default to speed auto and duplex auto , and that is the recommended setting for nearly all access ports.

When one side is configured manually and the other is left on auto, the auto side cannot negotiate:

Speed can still be detected by sensing the electrical signal (parallel detection), so speed normally matches.

Duplex cannot be sensed. IEEE rules say the auto side must assume half duplex at 10 or 100 Mbps, and full duplex at 1 Gbps and above.

So if a switch port is hard-coded to speed 100 and duplex full , and the PC is left on auto, the PC will choose 100 Mbps half duplex—a duplex mismatch. If speeds differ, for example one side forced to 100 and the other forced to 1000, the link simply stays down (down/down).

Figure 3-6 How a duplex mismatch arises when only one side is hard-coded.

SW1(config)# interface gigabitethernet1/0/5
SW1(config-if)# speed 100 ! options: 10 | 100 | 1000 | auto
SW1(config-if)# duplex full ! options: half | full | auto
SW1(config-if)# description PC-Finance-12
! Best practice: return the port to autonegotiation
SW1(config-if)# speed auto
SW1(config-if)# duplex auto

Tip

The golden rule: configure both ends for auto, or configure both ends manually with the same values. Never mix. On many Catalyst platforms, setting both speed and duplex manually disables autonegotiation on that port.

3.6 Reading show interfaces

The show interfaces command is your primary Layer 1 and Layer 2 diagnostic tool. The first line gives two status values: the line status (Layer 1: is there a signal?) and the protocol status (Layer 2: is the link usable?).

Table 3-5 Interface status combinations

Line status / protocol Meaning and typical cause
up / up (connected) Working normally
administratively down / down
(disabled)
Interface has the shutdown command

down / down (notconnect) No cable, cable fault, far end off or shut, speed mismatch up / down Layer 1 OK but Layer 2 problem: encapsulation mismatch, keepalive failure (mostly on routers/serial links)

down / down (err-disabled) Disabled by a feature such as port security or BPDU guard

Sw1 — ssh

SW1# show interfaces gigabitethernet1/0/5

GigabitEthernet1/0/5 is up, line protocol is up (connected)

Hardware is Gigabit Ethernet, address is 7069.5a3b.4c85 (bia 7069.5a3b.4c85) Description: PC-Finance-12

MTU 1500 bytes, BW 100000 Kbit/sec, DLY 100 usec,
reliability 251/255, txload 1/255, rxload 1/255
Encapsulation ARPA, loopback not set
Keepalive set (10 sec)
Full-duplex, 100Mb/s, media type is 10/100/1000BaseTX
input flow-control is off, output flow-control is unsupported
ARP type: ARPA, ARP Timeout 04:00:00
Last input 00:00:00, output 00:00:01, output hang never
Last clearing of "show interface" counters 01:12:44

Input queue: 0/2000/0/0 (size/max/drops/flushes); Total output drops: 0 Queueing strategy: fifo

Output queue: 0/40 (size/max)
5 minute input rate 412000 bits/sec, 61 packets/sec
5 minute output rate 1880000 bits/sec, 172 packets/sec
248211 packets input, 31520335 bytes, 0 no buffer
Received 1203 broadcasts (884 multicasts)
1734 runts, 0 giants, 0 throttles
2961 input errors, 2215 CRC, 0 frame, 0 overrun, 0 ignored
0 watchdog, 884 multicast, 0 pause input
0 input packets with dribble condition detected
661734 packets output, 812092245 bytes, 0 underruns
0 output errors, 0 collisions, 1 interface resets
0 unknown protocol drops
0 babbles, 0 late collision, 0 deferred
0 lost carrier, 0 no carrier, 0 pause output
0 output buffer failures, 0 output buffers swapped out

Figure 3-7 show interfaces on the full-duplex side of a duplex mismatch: rising runts, CRC and input errors. Output is illustrative of IOS 15/IOS XE.

Table 3-6 Key interface counters

Counter What it means Likely causes

Runts Frames smaller than 64 bytes Collisions (duplex mismatch), faulty NIC Giants Frames larger than the maximum (1518

A duplex mismatch is insidious because the link comes up and light traffic works; pings may even succeed. As load increases, the half-duplex side senses the full-duplex side's transmissions as collisions, stops sending and backs off. It records collisions and especially late collisions. Meanwhile, the full-duplex side receives fragments of frames cut off by the half-duplex side's jam signals and records runts, CRC errors and input errors. Throughput collapses. Use this process:

1. Run show interfaces status on the switch (or show interfaces ) and look at the duplex and speed columns. A prefix of a- (such as a-full ) means the value was autonegotiated; no prefix means it was configured.
2. Compare counters on both ends. Late collisions on one side plus CRC errors/runts on the other is the signature of a mismatch.
3. If the neighbour is a Cisco device, CDP reports the problem automatically (Figure 3-8).
4. Fix it by setting both ends to auto (or both to the same fixed values), then clear counters and confirm the errors stop increasing.

Sw1 — ssh

SW1# show interfaces status

Port Name Status Vlan Duplex Speed Type

Gi1/0/1 Uplink-SW2 connected trunk a-full a-1000 10/100/1000BaseTX Gi1/0/2 notconnect 1 auto auto 10/100/1000BaseTX

Gi1/0/3 SW3 connected 1 full 100 10/100/1000BaseTX
Gi1/0/5 PC-Finance-12 connected 10 full 100 10/100/1000BaseTX
Gi1/0/6 Printer-2F disabled 10 auto auto 10/100/1000BaseTX

%CDP-4-DUPLEX_MISMATCH: duplex mismatch discovered on GigabitEthernet1/0/3 (not half duplex), with SW3 FastEthernet0/1 (half duplex).

SW3# show interfaces fastethernet0/1 | include duplex|collision

Half-duplex, 100Mb/s, media type is 10/100BaseTX
0 output errors, 1931 collisions, 2 interface resets
0 babbles, 147 late collision, 0 deferred

Figure 3-8 Finding a duplex mismatch: hard-coded values on SW1 (no a- prefix), a CDP warning, and late collisions on SW3's half-duplex side.

Exam alert

Link the symptom to the cause: late collisions on a modern switched link almost always mean a duplex mismatch. CRC errors with no collisions anywhere point instead to cabling problems such as a damaged cable, EMI or a run longer than 100 m. A speed mismatch brings the link down entirely.

3.7 Physical Topologies and PoE

The physical topology describes how devices are cabled. Bus and ring topologies are historical; modern LANs use the following:

Star — every device connects to a central switch. A cable fault affects only one device. This is the standard access-layer design.

Extended star (hierarchical) — access switches connect to distribution switches, which connect to the core.

Full mesh — every device connects to every other; n(n−1)/2 links for n devices. Maximum redundancy, high cost; used between a few core devices and in spine-leaf designs (every leaf to every spine).

Partial mesh — some redundant links, chosen where they matter most; common in WANs.

Point-to-point — a single link between two devices, such as a WAN circuit.

Star Full mesh (n = 4, 6 links) Partial mesh

central switch, one cable per host n(n−1)/2 links redundancy where needed

Figure 3-9 Common physical topologies.

Chapter 1 introduced Power over Ethernet. PoE runs over the same Cat5e or better copper as data (it cannot run over fibre), within the 100 m limit. On a Catalyst switch, ports default to power inline auto , which detects and powers any standards-compliant device. power inline never disables PoE on a port, and power inline static max reserves power in advance. Verify with show power inline :

Sw1 — ssh

SW1# show power inline

--------- ------ ---------- ------- ------------------- ----- ---- Gi1/0/1 auto off 0.0 n/a n/a 30.0

Gi1/0/7 auto on 25.5 AIR-AP2802I-B-K9 4 30.0
Gi1/0/8 auto on 6.3 IP Phone 8845 2 30.0
Gi1/0/9 auto on 14.5 Ieee PD 3 30.0
Gi1/0/10 never off 0.0 n/a n/a 30.0

Figure 3-10 PoE budget and per-port power on a 370 W access switch.

LAB Create and fix a duplex mismatch

Goal: Observe the effect of hard-coding one side of a link, then restore correct operation. Topology: SW1 Gi1/0/3 connected to SW3 Fa0/1 (or Gi0/1) with a straight-through cable (Auto-MDIX handles the crossover). Both switches start with default port settings. Steps:

1. Verify the starting state: show interfaces status on both switches should show a-full a-100 (or a-1000 ).
2. Hard-code SW1's side only:

SW1(config)# interface gigabitethernet1/0/3
SW1(config-if)# speed 100
SW1(config-if)# duplex full

3. On SW3, check show interfaces fa0/1 . The port now shows Half-duplex, 100Mb/s . Watch for the CDP duplex-mismatch log message on both switches.
4. Generate traffic (for example, extended pings with large sizes and repeat counts between management SVIs) and run show interfaces on both sides. Note collisions/late collisions on SW3 and CRC/runts on SW1.
5. Fix it by restoring autonegotiation on SW1, then clear the counters:

SW1(config-if)# speed auto
SW1(config-if)# duplex auto
SW1(config-if)# end
SW1# clear counters gigabitethernet1/0/3

Verify: show interfaces status shows a-full on both ends, and repeated show interfaces output shows CRC, runts and late-collision counters remaining at 0.

Expected result: Both sides negotiate full duplex at the highest common speed, the CDP warning stops, and throughput returns to normal.

Chapter review

Key terms

Csma/cd

collision domain
half duplex
full duplex

Utp

Sfp / sfp+

MAC address

Oui

1. What is the maximum segment length for 1000BASE-T over Cat5e UTP?

A. 55 m
B. 100 m
C. 185 m
D. 550 m

2. Which two statements describe single-mode fibre compared with multimode fibre? (Choose two.)

A. It has a larger core diameter
B. It typically uses a laser light source
C. It supports longer distances
D. It is limited to 100 m
E. It uses LED light sources exclusively

3. Which cable type is required to connect a PC directly to a router's Ethernet port when Auto-MDIX is not available?

A. Straight-through
B. Crossover
C. Rollover
D. Single-mode fibre

4. On which pins does a PC NIC transmit when using 100BASE-TX?

A. 1 and 2
B. 3 and 6
C. 4 and 5
D. 7 and 8

5. A switch port is configured with speed 100 and duplex full . The attached PC uses autonegotiation. What will the PC choose?

A. 100 Mbps, full duplex
B. 100 Mbps, half duplex
C. 1000 Mbps, full duplex
D. The link will stay down

6. An interface shows a steadily increasing number of late collisions. What is the most likely cause?

A. A speed mismatch
B. A duplex mismatch
C. An MTU mismatch
D. A VLAN mismatch

7. Which counter increments when a received frame is smaller than 64 bytes?

A. Giants
B. Runts
C. Late collisions
D. Output errors

8. Which two conditions commonly cause CRC errors? (Choose two.)

A. Electromagnetic interference on a UTP run
B. An interface in the shutdown state
C. A duplex mismatch
D. A missing default gateway
E. An incorrect VLAN assignment

9. What does the OUI portion of a MAC address identify?

A. The VLAN
B. The manufacturer
C. The switch port
D. Whether the address is broadcast

10. Which destination MAC address is used by an Ethernet broadcast?

A. 0000.0000.0000
B. 0100.5E00.0001
C. FFFF.FFFF.FFFF
D. 3333.0000.0001

11. In show interfaces status , what does a-full in the Duplex column indicate?

A. Full duplex was configured manually
B. Full duplex was autonegotiated
C. The port is administratively down
D. The port uses Auto-MDIX

12. Which connector is typically used with SFP modules on Cisco switches?

A. RJ-45
B. LC
C. SC
D. ST

13. How many links are needed to build a full mesh between five routers?

A. 5
B. 8
C. 10
D. 20

Answers

1. B. All BASE-T standards share the 100 m limit. 55 m is the limit for 10GBASE-T on Cat6; 185 m was 10BASE2 coax; 550 m is a multimode fibre distance.
2. B, C. SMF uses a narrow (~9 µm) core with laser sources and reaches kilometres. MMF has the larger core and uses LED/VCSEL sources.
3. B. A PC and a router both transmit on pins 1,2, so they are “like” devices and need a crossover cable. Rollover cables are console cables.
4. A. PCs, routers and APs transmit on 1,2 and receive on 3,6; switches do the opposite.
5. B. The PC senses the speed (100 Mbps) but, receiving no negotiation, defaults to half duplex at 10/100 Mbps, causing a duplex mismatch.
6. B. Late collisions on a switched link are the classic symptom of a duplex mismatch. A speed mismatch takes the link down.
7. B. Runts are frames under 64 bytes; giants exceed the maximum size.
8. A, C. EMI corrupts bits, and a duplex mismatch truncates frames; both cause FCS failures. The other options do not produce CRC errors.
9. B. The Organizationally Unique Identifier is the first 24 bits and identifies the vendor.
10. C. FFFF.FFFF.FFFF is broadcast. 0100.5Exx.xxxx is IPv4 multicast; 3333.xxxx.xxxx is IPv6 multicast.
11. B. The a- prefix means the value was autonegotiated.
12. B. SFP and SFP+ fibre modules use duplex LC connectors.
13. C. n(n−1)/2 = 5 × 4 / 2 = 10 links.