Sessions 6,7 Ethernet

Ethernet Fundamentals
3
Part 1
 Introduction to Ethernet
Part 2
 Layer 2 and Ethernet Switches
 Cables, Duplex, and Troubleshooting
 Ethernet and the OSI Model – more detail
 Ethernet frames – more detail
Soft Polynomials (I) Pvt. Ltd. Subhash Iyer, Program Head

Ethernet Local Area Networks (LANs)
5
 LAN (Local Area Network) - A computer network connected through a
wired or wireless medium by networking devices (hubs, switches, routers)
and administered by a single organization.
 Ethernet – A family of Layer 2 Data Link protocols for Local Area Networks
.

IEEE Standards
6
Brief History:
 1970’s - Robert
Metcalfe and his
coworkers at Xerox
PARC
 1980 - Ethernet
protocol published
by Digital
Equipment
Corporation, Intel,
and Xerox (DIX)
 1985 - Institute of
Electrical and
Electronics
Engineers (IEEE)
published IEEE
802.2 and 802.3

Our Focus is Ethernet
 History
 Roots in Aloha packet-radio network
 LAN standards define MAC and physical layer connectivity
 IEEE 802.3 (CSMA/CD - Ethernet) standard – originally 2Mbps
 IEEE 802.3u standard for 100Mbps Ethernet
 IEEE 802.3z standard for 1,000Mbps Ethernet
 CSMA/CD: Ethernet’s Media Access Control (MAC) policy
 CS = carrier sense
 Send only if medium is idle
 MA = multiple access
 CD = collision detection
 Stop sending immediately if collision is detected
7

8
LLC (Logical Link Control)
MAC (Media Access Control)
IEEE 802 Extension to the
OSI Model
 The Institute of Electrical and Electronic Engineers (IEEE) is a
professional organization that defines network standards.
 IEEE 802.3 “Ethernet” is the predominant and best known LAN
standards, along with 802.11 (WLAN).
 The IEEE divides the OSI data link layer into two separate sublayers.
Recognized IEEE sublayers are:
 Media Access Control (MAC) (transitions down to media)
 Logical Link Control (LLC) (transitions up to the network layer)
Data Link Sublayers

Refresher: OSI Model
 Open Systems Interconnection model (OSI):
a conceptual model that characterizes and
standardizes the internal functions of
a communication system by partitioning it
into abstraction layers.
 The model is a product of the Open Systems
Interconnection project at the International
Organization for Standardization (ISO), maintained
by the identification ISO/IEC 7498-1.
9

 The model groups communication
functions into seven logical layers.
 A layer serves the layer above it
and is served by the layer below it.
 For example, a layer that provides
error-free communications across
a network provides the path
needed by applications above it,
while it calls the next lower layer to
send and receive packets that make
up the contents of that path.
 Two instances at one layer are
connected by a horizontal
connection on that layer.
10

11

Ethernet Standard Defines Physical Layer
 802.3 standard defines both MAC and physical
layer details
12
Metcalfe’s original
Ethernet Sketch

13
LLC (Logical Link Control)
MAC (Media Access Control)
IEEE 802 Extension to the
OSI Model
 The Institute of Electrical and Electronic Engineers (IEEE) is a
professional organization that defines network standards.
 IEEE 802.3 “Ethernet” is the predominant and best known LAN
standards, along with 802.11 (WLAN).
 The IEEE divides the OSI data link layer into two separate sublayers.
Recognized IEEE sublayers are:
 Media Access Control (MAC) (transitions down to media)
 Logical Link Control (LLC) (transitions up to the network layer)
Again: Data Link Sublayers

LLC – Logical Link Sublayer
14
 Logical Link Control (LLC) defined in the IEEE 802.2 specification
 Provides versatility in services to network layer protocols that are above
it, while communicating effectively with the variety of technologies below
it.
 The LLC, as a sublayer, participates in the encapsulation process.

802.2 LLC
15
IPX IP APPLE-
TALK
LLC
Layer 3
Layer 2 - LLC
MAC &Layer 1 Ethernet Token
Ring
FDDI
* *
**
* Legacy technologiesSoft Polynomials (I) Pvt. Ltd. Subhash Iyer, Program Head

A Quick Word about Token Ring
 Developed by IBM in early 80’s as a new LAN
architecture
 Consists of nodes connected into a ring (typically via
concentrators)
 Special message called a token is passed around the ring
 When nodes gets the token it can transmit for a limited time
 Every node gets an equal opportunity to send
 IEEE 802.5 standard for Token Ring
 Designed for predictability, fairness and reliability
 Originally designed to run at either 4Mbps and 16Mbps
 Still used and sold but beaten out by Ethernet
16

17
Application
Header + data
802.2 LLC Data Encapsulation Example
We will been focusing on the Layer 2, Data Link, Ethernet
Frame for now.
010010100100100100111010010001101000…
Application Layer
Layer 4: Transport Layer
Layer 3: Network Layer
Layer 2:
Data Link
Layer
Layer 1: Physical
Layer

Ethernet Technologies: 10Base2
 10: 10Mbps; 2: under 185 (~200) meters cable length
 Thin coaxial cable in a bus topology
 Repeaters used to connect multiple segments
 Repeater repeats bits it hears on one interface to its other interfaces:
physical layer device only!
18

10BaseT and 100BaseT
 10/100 Mbps rate
 T stands for Twisted Pair
 Hub(s) connected by twisted pair facilitate “star topology”
 Distance of any node to hub must be < 100M
19
Location 1 Location 2 Location3

Physical Layer Configurations for 802.3
 Physical layer configurations are specified in three parts
 Data rate (10, 100, 1,000)
 10, 100, 1,000Mbps
 Signaling method (base, broad)
 Baseband
 Digital signaling
 Broadband
 Analog signaling
 Cabling (2, 5, T, F, S, L)
 5 - Thick coax (original Ethernet cabling)
 F – Optical fiber
 S – Short wave laser over multimode fiber
 L – Long wave laser over single mode fiber
20

Ethernet Overview
 Most popular packet-switched LAN technology
 Bandwidths: 10Mbps, 100Mbps, 1Gbps
 Max bus length: 2500m
 500m segments with 4 repeaters
 Bus and Star topologies are used to connect hosts
 Hosts attach to network via Ethernet transceiver or hub or switch
 Detects line state and sends/receives signals
 Hubs are used to facilitate shared connections
 All hosts on an Ethernet are competing for access to the medium
 Switches break this model
 Problem: Distributed algorithm that provides fair access
21

Ethernet Overview (contd.)
 Ethernet by definition is a broadcast protocol
 Any signal can be received by all hosts
 Switching enables individual hosts to communicate
 Network layer packets are transmitted over an
Ethernet by encapsulating
 Frame Format
22
Dest
addr
64 48 32
CRCPreamble Src
addr
Type Body
1648

Ethernet Frames
 Preamble is a sequence of 7 bytes,
each set to “10101010”
 Used to synchronize receiver before
actual data is sent
 Addresses
 unique, 48-bit unicast address
assigned to each adapter
 example: 8:0:e4:b1:2:af
 Each manufacturer gets their own
address range
 broadcast: all 1s
 multicast: first bit is 1
23
• Type field is a demultiplexing key
used to determine which higher
level protocol the frame should be
delivered to
• Body can contain up to 1500 bytes
of data
Soft Polynomials (I) Pvt. Ltd.

A Quick Word about Aloha Networks
 Developed in late 60’s by Norm Abramson at Univ. of
Hawaii (!!) for use with packet radio systems
 Any station can send data at any time
 Receiver sends an ACK for data
 Timeout for ACK signals that there was a collision
 What happens if timeout is poorly timed?
 If there is a collision, sender will resend data after a random backoff
 Utilization (fraction of transmitted frames avoiding
collision for N nodes) was pretty bad
 Max utilization = 18%
 Slotted Aloha (dividing transmit time into windows) helped
 Max utilization increased to 36%
24

Ethernet’s MAC Algorithm
 In Aloha, decisions to transmit are made without paying
attention to what other nodes might be doing
 Ethernet uses CSMA/CD – listens to line before/during
sending
 If line is idle (no carrier sensed)
 send packet immediately
 upper bound message size of 1500 bytes
 must wait 9.6us between back-to-back frames
 If line is busy (carrier sensed)
 wait until idle and transmit packet immediately
 called 1-persistent sending
 If collision detected
 Stop sending and jam signal
 Try again later
25

Collisions
 Collisions are caused when two adaptors transmit
at the same time (adaptors sense collision based on
voltage differences)
• Both found line to be idle
• Both had been waiting to for a busy line to become idle
26
A B
A B
A starts at
time 0
Message almost
there at time T when
B starts – collision!
How can we be sure A knows about the collision?

Collision Detection
 How can A know that a collision has taken place?
 There must be a mechanism to insure retransmission on collision
 A’s message reaches B at time T
 B’s message reaches A at time 2T
 So, A must still be transmitting at 2T
 IEEE 802.3 specifies max value of 2T to be 51.2us
 This relates to maximum distance of 2500m between hosts
 At 10Mbps it takes 0.1us to transmit one bit so 512 bits (64B) take 51.2us to
send
 So, Ethernet frames must be at least 64B long
 14B header, 46B data, 4B CRC
 Padding is used if data is less than 46B
 Send jamming signal after collision is detected to insure all hosts see
collision
 48 bit signal
27

Collision Detection contd.
 Video
28

State Diagram for CSMA/CD
29
Packet?
Sense
Carrier
Discard
Packet
Send Detect
Collision
Jam channel
b=CalcBackoff();
wait(b);
attempts++;
No
attempts < 16
attempts == 16
Yes
NoYes

Exponential Backoff
 If a collision is detected, delay and try again
 Delay time is selected using binary exponential backoff
 1st time: choose K from {0,1} then delay = K * 51.2us
 2nd time: choose K from {0,1,2,3} then delay = K * 51.2us
 nth time: delay = K x 51.2us, for K=0..2n – 1
 Note max value for k = 1023
 give up after several tries (usually 16)
 Report transmit error to host
 If delay were not random, then there is a chance that
sources would retransmit in lock step
 Why not just choose from small set for K
 This works fine for a small number of hosts
 Large number of nodes would result in more collisions
30

CSMA/CD Operation
31

IEEE 802.3 CSMA/CD Flow Chart
32

Ethernet is Best Effort Delivery
33
 Ethernet is best-
effort delivery, no
guarantee.
 Like a trucking
service, it doesn’t
really know or
care about the
what it is carrying.

The IEEE Working Groups
34
802.1
802.2
802.3
802.4
802.5
802.6
802.7
802.8
802.9
Networking Overview and Architecture
Logical Link Control
Ethernet
Token Bus
Token Ring
MANs
Broadband
Fiber Optic
Isochronous LAN
...and more!
802.11 Wireless LAN

IEEE Identifiers
35
 3 part identifier
 Speed in Mbps
 Type of signaling used (Baseband or Broadband)
 Distance or Medium
 Early days: Cable Distance in meters, rounded to the nearest 100
meters
 Later days: Physical medium used
Early Standards Older Fiber
Standards
100 Mbps Media 1000 Mbps Media
10BASE5 10BASE-F 100BASE-T 1000BASE-X
10BASE2 10BASE-FB 100BASE-X 1000BASE-SX
FOIRL 10BASE-FP 100BASE-TX 1000BASE-LX
10BROAD36 10BASE-FL 100BASE-FX 1000BASE-CX
1BASE5 100BASE-T4 1000BASE-T
10BASE-T 100BASE-T2
Many of these standards were short lived or never implemented

IEEE Identifiers
37
 10BASE5 (Thick Ethernet)
 10 refers to 10 Mbps
 Baseband: Dedicated to carrying one type of service
 Broadband: (Cable television) Designed to deliver multiple channels
 5 refers to 500 meter maximum distance
 100BASE-TX (Most widely used variety of Fast Ethernet)
 100 refers to 100 Mbps
 TX Two pairs of Category 5 Twisted-pair cable

Network Interface Card (NIC)
38

39
 Layer 2, Data Link Layer, device
 Connects the device (computer) to
the LAN
 Responsible for the local Layer 2
address (later)
 Common Layer 2 NICs:
 Ethernet
 Token Ring
 Common Bandwidth
 10 Mbps, 10/100 Mbps,
10/100/1000 Mbps

Tracing the Physical Connection
NIC (Network Interface Card)
40

Connecting the NIC to a Hub or Switch…
41

From PC to Ethernet Port…
42

From Ethernet Port to Patch Panel…
43
Back View Front View

From Patch Panel to Switch (or hub)
44

From PC to Switch
45

All of that is the same as these!
46

Our focus!
47
 Ethernet protocol is only concerned with how the
information gets from one Ethernet host or device
to another.
 This is done at the MAC Sub-Layer

Media Access Control Protocol
49
 Original Ethernet standard based on CSMA/CD media access control (MAC)
 Also known as Half-duplex mode
 No need for CSMA/CD in Full-duplex mode (later)
 Compete for a shared Ethernet channel in a fair and equitable manner

Generic Data Link Frame Format
50
Preamble or Start Field
 When computers are connected to a physical medium, there must be a way
they can grab the attention of other computers to broadcast the message,
"Here comes a frame!"
 Various technologies have different ways of doing this process, but all
frames, regardless of technology, have a beginning signaling sequence of
bytes.
 Depending upon frame format: Preamble = 7 bytes, Start or Start of Frame
Delimiter (SFD) = 1 byte

51
Address Field
 We saw how IEEE 802.3 uses Destination and Source Addresses.
 Ethernet:
 Unicast address – MAC address of a single device
 Broadcast address – All devices (All 1 bits, All F’s in Hex)
 Multicast address – Specific group of devices

Unicast, Multicast, Broadcast
Destination Addresses
52
 Unicast address: A single Ethernet frame to be received by a single
station.
 Unknown Unicast: This is from the perspective of a switch, when the
unicast address is not in its MAC Address Table
 Multicast address: A single Ethernet frame to be received by a group of
stations.
 Broadcast address: Special case of a multicast address, which is all 1’s.
This is an Ethernet frame to be received by all stations.

53
Type Field
 Usually information indicating the layer 3 protocols in the data field, I.e. IP
Packet.
 Type field values of particular note for IEEE 802.3 frames include:
 0x0600 XNS (Xerox)
 0x0800 IP (Internet protocol)
 0x8137 Novell NetWare packet formatted for Ethernet II
 0x0806 ARP Message

54
Data Field
 Included along with this data, you must also send a few other bytes.
 They are called padding bytes, and are sometimes added so that the frames
have a minimum length for timing purposes.
 LLC bytes are also included with the data field in the IEEE standard frames.
(later)

55
 The fields of various Ethernet framing that are used for identifying the type of data
contained in a frame:
 Ethernet II or DIX (DEC, Intel, Xerox) – Most common
 IEEE Ethernet (802.3)
 IEEE 802.3 with SNAP header

IFG – Interframe Gap
56
 Ethernet devices must allow a minimum idle period between transmission of
frames known as the interframe gap (IFG) or interpacket gap (IPG).
 Note: Both half and full-duplex
 It provides a brief recovery time between frames to allow devices to prepare
for reception of the next frame.
 The minimum interframe gap is:
 10 Mbps Ethernet: 96 bit times, which is 9.6 microseconds (millionths
of a second)
 100 Mbps, Fast Ethernet: 960 nanoseconds (billionths of a second)
 1000 Mbps, Gigabit Ethernet: 96 nanoseconds
 Note: 802.11 (WLAN) uses similar
Ethernet
Frame
IFG Ethernet
Frame
IFG Ethernet
Frame
IFG Ethernet
Frame
IFG

Collisions, Slot time and Minimum Frame Size
57
Notes
 Original Ethernet (802.3) designed as Half-duplex
 CSMA/CD is based on half-duplex and is NOT part of full-duplex
 Collisions are part of CSMA/CD and half-duplex Ethernet
 Collisions are a normal part of operation and are NOT errors
 Collisions are NOT part of full-duplex Ethernet

Collision Domain
58
 Collision Domain: Refers to a single half-duplex Ethernet system whose
elements (cables, repeaters, hubs, station interfaces and other network
hardware) are all part of the same signal timing domain.
 If two or more devices transmit at the same time a collision will occur.
 If a collision is detected, the station will continue to transmit 32 bits called
the collision enforcement jam signal.

Collision Domain
59
 Switches do not forward collision signals

Slot Time and Maximum Cable Length
60
 Slot time
 Time it takes for a signal to travel from one end of the maximum-sized
system to the other end and return (round trip propagation time) within a
collision domain.
 Maximum time required by collision enforcement.
 After this amount of time (or bits), device assumes no collision.
 Ethernet and Fast Ethernet
 Slot time = 512 bit times (the time it takes to transfer 512 bits)
If a collision occurs it will
be within the first 512
bits that I send.
bits that I send.

61
 Slot time and maximum cable length are tightly coupled.
 Original 10 Mbps Ethernet: On coaxial cable, signals could travel 2,800
meters (9,186 feet) and back in 512 bit times.
 Maximum distance of collision domain is 2,800 meters.
 In other words, a station would know about a collision (rise in DC signal
level) before it transmitted the 513th bit.
 Fast Ethernet Twisted-pair maximum network diameter is 205 meters or
672 feet, but is limited by cabling standards of 100 meters or 328 feet.
(Remember, more bits per second, shorter bits, than Ethernet)
bits that I send.
bits that I send.

62
 512 bit Slot Time
 Destination Address = 48 bits
 Source Address = 48 bits
 Type = 16 bits
 Data = 368 bits or (46 bytes * 8 bits per byte)
 FCS = 32 bits
 This is why there is a minimum of 46 bytes of data!
If a collision occurs
it will be within the
first 512 bits that I
send.
512 bit minimum

63
 A collision will be noticed within the first 512 bits transferred, so the minimum
frame size must be 512 bits.
 After 512 bits, the sending station assumes no collisions.
 At 513 bits, all stations on the entire Ethernet system, collision domain (cable,
repeaters, hubs) should have seen this frame by now before they begin
transmitting.
 This is why there is a maximum size to the Ethernet system. (Half-duplex
only!)
If a collision occurs
it will be within the
first 512 bits that I
send.
512 bit minimum

BREAK KE BAAD!!!!
64
Subhash Iyer, Program Head

MAC – Media Access Control Sublayer
2
 The Media Access Control (MAC) sublayer deals with the protocols
that a host follows in order to access the physical media.
 Defined in IEEE 802.3 specification
 Responsible for the actual framing
 Builds the 1s and 0s to hand off to the physical layer.
 Responsible for media access (CSMA/CD)

Ethernet and IEEE 802.3
3
 This standard includes the protocol used to “frame” the data by the
sending Ethernet host computer.
 Most of the time, the term “Ethernet” is used to mean IEEE 802.3
 Ethernet and IEEE 802.3 are used interchangeably, even though they
are not really the same thing. (more later)

Ethernet “Data”
4
Data may be:
 IP Packet
 ARP Message
 Other

The MAC Address
5
 Part of the Ethernet protocol includes the MAC
(Media Access Control)
 Every Ethernet NIC card has a unique MAC address.
 MAC addresses provide a way for computers to
identify themselves.
 They give hosts a permanent, unique name.

The MAC Address
6
 MAC addresses are:
 48 bits in length
 Expressed as 12 hexadecimal digits.
 The first 6 hexadecimal digits, which are administered by the IEEE,
identify the manufacturer or vendor and thus comprise the Organizational
Unique Identifier (OUI).
 The remaining 6 hexadecimal digits comprise the interface serial
number, or another value administered by the specific vendor.
 MAC addresses are sometimes referred to as burned-in addresses (BIAs)
because they are burned into read-only memory (ROM) and are copied into
random-access memory (RAM) when the NIC initializesSoft Polynomials (I) Pvt. Ltd. Subhash Iyer, Program Head

The MAC Address
7
 The Ethernet protocol uses MAC addresses to identify the source of the
Ethernet frame and the destination of the Ethernet frame.
 Whenever a computer sends an Ethernet frame, it includes the MAC address on
its NIC as the Source “MAC” Address.
 We will learn later how it learns the Destination “MAC” Address.
 We will see how all of this works in a moment.
MAC
Address
MAC
Address

MAC Address Format
8
OUI unique
 An Intel MAC address: 00-20-E0-6B-17-62
 0000 0000 - 0010 0000 – 1110 0000 - 0110 1011 – 0001 0111 – 0110 0010
 IEEE OUI FAQs: http://standards.ieee.org/faqs/OUI.html
Dec Bin Hex Dec Bin Hex
0 = 0000 = 0 8 = 1000 = 8
1 = 0001 = 1 9 = 1001 = 9
2 = 0010 = 2 10 = 1010 = A
3 = 0011 = 3 11 = 1011 = B
4 = 0100 = 4 12 = 1100 = C
5 = 0101 = 5 13 = 1101 = D
6 = 0110 = 6 14 = 1110 = E
7 = 0111 = 7 15 = 1111 = F

What is the Address on my NIC?
9

MAC Addresses Are Flat
10
 MAC addresses provide a way for computers to identify themselves.
 They give hosts a permanent, unique name.
 The number of possible MAC addresses is 16^12 (or over 2 trillion!).
 MAC addresses do have one major disadvantage:
 They have no structure, and is considered flat address space.
 Like using just a name when sending a letter instead of a structured address.

Generic Data Link Frame
11
 A message is “framed” (layer 2) and transmitted on the
cable (layer 1) by the Ethernet NIC.
 Framing provides order, or structure, to the stream of
bits, bitstream.

Bringing it all together…
12
 Let’s pause here for a moment and figure all of this
out!
 Let’s bring the following together:
 Ethernet Frames and MAC Addresses
 Sending and receiving Ethernet frames on a bus
 CSMA/CD
 Sending and receiving Ethernet frames via a hub
 Sending and receiving Ethernet frames via a switch

Serial vs Multiaccess Network
13Serial
Multiaccess

14

Ethernet: Multiaccess Network
15

Bus Topology
16
A bus topology uses a single backbone segment (length of
cable) that all the hosts connect to directly.
Original Ethernet used a bus topology.
By the way, Ethernet hubs work the same as a “bus”.

Sending and receiving Ethernet frames on a bus
17
 When an Ethernet frame is sent out on the “bus”
all devices on the bus receive it.
 What do they do with it?
1111 2222 3333 nnnn Abbreviated
MAC
Addresses
11113333

18
 When information (frame) is transmitted, every PC/NIC on the shared media
copies part of the transmitted frame to see if the destination address
matches the address of the NIC.
 If there is a match, the rest of the frame is copied
 If there is NOT a match the rest of the frame is ignored.
 Unless you are running a protocol analyzer program such as Ethereal.
MAC
Addresses
11113333
Nope Nope
Hey, that’s
me!

19
 So, what happens when multiple computers try to
transmit at the same time?
MAC
Addresses

20
Collision!
MAC
Addresses
X

Access Methods
21
Two common types of access methods for LANs include
 Non-Deterministic: Contention methods (Ethernet, IEEE
802.3)
 Only one signal can be on a network segment at
one time.
 Collisions are a normal occurrence on an
Ethernet/802.3 LAN
 Deterministic: Token Passing (Token Ring)

CSMA/CD (Carrier Sense Multiple Access
with Collision Detection)
22
CSMA/CD Common contention method used with Ethernet and
IEEE 802.3
 “Let everyone have access whenever they want and we will work it
out somehow.”

CSMA/CD and Collisions
23
CSMA/CD (Carrier Sense Multiple Access with Collision Detection)
 Listens to the network’s shared media to see if any other users on “on the
line” by trying to sense a neutral electrical signal or carrier.
 If no transmission is sensed, then multiple access allows anyone onto
the media without any further permission required.
 If two PCs detect a neutral signal and access the shared media at the exact
same time, a collision occurs and is detected.
 The PCs sense the collision by being unable to deliver the entire frame onto
the network. (This is why there are minimum frame lengths along with
cable distance and speed limitations. This includes the 5-4-3 rule.)
 When a collision occurs, a jamming signal is sent out by the first PC to
detect the collision.
 Using either a priority or random backoff scheme, the PCs wait
certain amount of time before retransmitting.
 If collisions continue to occur, the PCs random interval is doubled,
lessening the chances of a collision.

CSMA/CD and Collisions
24
And as we said,
 When information (frame) is transmitted, every PC/NIC on the shared
media copies part of the transmitted frame to see if the destination
address matches the address of the NIC.
 If there is a match, the rest of the frame is copied
 If there is NOT a match the rest of the frame is ignored.
MAC
Addresses
11113333
Nope Nope
Hey, that’s
me!
Notice the
location of
the DA!

Sending and receiving Ethernet frames via a hub
25
 Only one device on
the hub can
communicate at a
time, otherwise
collisions occur.
 10 Mbps ports are the
most common.
 100/1000 Mbps also
“available”.
 The hub acts the same
as a “bus”.
Hub or

26
 So, what does a hub
do when it receives
information?
 A hub is nothing
more than a
multiport repeater.
1111 2222
3333 4444
5555
?
11113333

Repeaters
27
 Signals can only travel so far through media before
they weaken, and become garbled.
 This weakening of signals is called attenuation.
 Attenuation increases when:
 Media distances are lengthened
 Nodes are added to the media
 Repeaters:
 take in weakened signals
 clean them up
 regenerate them
 send them on their way along the network

Repeater: Layer 1 Device
28
 Repeaters are Layer 1 devices.
 They do NOT look at:
 Layer 2, Data Link (MAC, Ethernet)
addresses
 Layer 3, IP Addresses.
Signal come in
… signal go out.
(after I amplify
it)

Hub
29
 Hub is nothing but a multiport repeater.
 Hubs are Layer 1 devices.
 Data that comes in one port is sent out all other ports, except for the port
it came in on.

Hubs
30
 Hubs allow computers and other network devices to
communicate with each other, and use a star topology.
 Like a repeater, a hub regenerates the signal.
 Hubs have the same disadvantage as a repeater, anything it
receives on one port, it FLOODS out all other ports.
 Wherever possible, hubs should be replace by switches.

31
 The hub will flood it out all ports
except for the incoming port.
 Hub is a layer 1 device.
 A hub does NOT look at layer 2
addresses, so it is fast in
transmitting data.
 Disadvantage with hubs: A hub
or series of hubs is a single
collision domain (coming)
 A collision will occur if any two or
more devices transmit at the same
time within the collision domain.
1111 2222
3333 4444
5555
11113333
Nope
Nope
Nope
For me!Soft Polynomials (I) Pvt. Ltd. Subhash Iyer, Program Head

32
 Another disadvantage with hubs
is that is take up unnecessary
bandwidth on other links.
1111 2222
3333 4444
5555
11112222
Nope Nope
Nope
For me!
Wasted
bandwidth

33
 What happens when
two hosts on the
same hub, or when
multiple hubs are
connected, transmit
at the same time?
1111 2222
3333 4444
5555
?
11112222
33334444

34
 Collision occurs.
 Although, hubs have
little latency,
CSMA/CD requires
resending of frames
and adds latency.
1111 2222
3333 4444
5555
Collision
11112222
33334444
X

Half-duplex (Introduction)
35
 Hubs operate only in Half-duplex.
 Half-duplex means that only one end can send at a time.
 With half-duplex NICs, a host can only transmit or receive, not both at the same
time, or a collision will occur.
 When multiple devices are connected to a hub or series of hubs, only one device
can transmit.
 Uses CSMA/CD.
 If the a carrier is detected, then the NIC will not transmit.
 Ethernet hubs and repeaters can only operate in half-duplex mode.
Half-duplex

Half-Duplex mode
36
 All of these Ethernet
NICs and ports on the
hubs are operating in
Half-Duplex mode.
 When multiple devices
are connected to a hub
or series of hubs, only
one device can
transmit.

Collision Domain: Shared Access
37
 Collision domain (Wikipedia): A group of
Ethernet or Fast Ethernet devices in a
CSMA/CD LAN that are connected by
repeaters/hubs and compete for access on the
network.
 Only one device in the collision domain may
transmit at any one time, and the other
devices in the domain listen to the network
in order to avoid data collisions.
 A collision domain is sometimes referred to
as an Ethernet segment.
 If you connect several computers to a single
medium that is only connected by repeaters
and hubs (Layer 1 devices), you have a
shared-access situation, and you have a single
collision domain.

Full-duplex
38
 Full-duplex is allows simultaneous communication between a pair of stations
or devices.
 Full-duplex allows devices to send and receive at the same time.
 Both ends of the link must be in full-duplex mode.
 In full-duplex, the station ignores any collision detect signals that come from
the transceiver.
 If a hub is connected to a switch, the switch port must be in half-duplex.
 The collision domain will end at the switch port.Soft Polynomials (I) Pvt. Ltd. Subhash Iyer, Program Head

Where are the collision domains?
What would be the duplex settings?
39
hub hub hub hub hub hub
hub hub
router

40
hub hub
router
Single Collision Domain

41
hub hub
router
Half-duplex
Half-duplex
hub

42
switch switch
router

Switched Ethernet
 Switches forward and filter frames based on LAN addresses
 It’s not a bus or a router (although simple forwarding tables are
maintained)
 Very scalable
 Options for many interfaces
 Full duplex operation (send/receive frames simultaneously)
 Connect two or more “segments” by copying data frames
between them
 Switches only copy data when needed
 key difference from repeaters
 Higher link bandwidth
 Collisions are completely avoided
 Much greater aggregate bandwidth
 Separate segments can send at once
43

Sending and receiving Ethernet frames via a switch
44
 Layer 2 device (also includes
layer 1) which examines and
bases its decisions on the
information in layer 2 frames
 Switch ports typically operate in
full-duplex.
 Multiple devices on the switch
can communicate at a time,
otherwise collisions occur.
 10/100 Mbps ports are the most
common.
 1000 Mbps also are also
common, usually connecting to
another switch or router.

45
switch switch
router
Collision Domains Collision Domains

46
switch switch
router
Half-duplex
Half-duplex
Full-duplex
hub

47
switch hub hub switch switch switch
switch switch
router

48
switch switch
router
Collision Domains

49
switch switch
router
Full-duplex
Half-duplex
Full-duplex
switch

All scenarios are
multiaccess networks
50
switch switch
router

52
11 22 33 44 55 66
Hub Hub
Half-duplex
CSMA/CD

53
11 22 33 44 55 66
Switch Switch
MAC Address Table
Port Source MAC Add. Port Source MAC Add.
MAC Address Table
Full-duplex
No CSMA/CD

Full-duplex
54
 Full-duplex is an optional mode of operation allowing simultaneous
communication between a pair of stations or devices.
 Specified in IEEE 802.3x in March 1997

Full-duplex
55
 Full-duplex Ethernet allows the transmission of a packet and the reception
of a different packet at the same time.
 The full-duplex Ethernet switch takes advantage of the two pairs of wires in
the cable by creating a direct connection between the transmit (TX) at one
end of the circuit and the receive (RX) at the other end.
 Half Duplex Ethernet usually can only use 50%-60% of the available 10
Mbps of bandwidth because of collisions and latency.
 Full-duplex Ethernet offers 100% of the bandwidth in both
directions.
 10 Mbps Ethernet: This produces a potential 20 Mbps throughput, which
results from 10 Mbps TX and 10 Mbps RX.

Full-duplex
56
 IEEE 802.3x full-duplex
standard requires:
 The medium must have
independent transmit and
receive data paths that can
operate simultaneously.
 There are exactly two
stations connected with a full-
duplex point-to-point link.
 There is no CSMA/CD
multiple access algorithm,
since there is no contention
for a shared medium.
 Both stations on the LAN are
capable of, and have been
configured to use, the full-
duplex mode of operation.
 Ethernet hubs and repeaters
can only operate in half-duplex
mode.

Half-duplex Controller
57
 With half-duplex NICs,a host can only transmit or receive.
 If the a carrier is detected, then the NIC will not transmit.
 In full-duplex the station ignores the carrier sense and does not defer to
traffic being received on the channel.
 In full-duplex, the station ignores any collision detect signals that come from
the transceiver.
 Ethernet hubs and repeaters can only operate in half-duplex mode.
Half-duplex
controllers

Full-duplex
58
 Both half-duplex and full-duplex Ethernet uses an interframe gap (IFG).
 Full-duplex uses the IFG to ensure that the interfaces at each end of the link
can keep up with the full frame rate of the link.
 CSMA/CD not used in full-duplex Ethernet:
 No CS (Carrier Sense) – In full-duplex the station ignores carrier
sense since it can send whenever it likes.
 No MA (Multiple Access) – Since there is only one station at the other
end of the link and the Ethernet channel between them is not the subject
of access contention.
 No CD (Collision Detect) – Since there is no access contention, there
will be no collisions, and station can ignore CD.
Ethernet
Frame
IFG Ethernet
Frame
IFG Ethernet
Frame
IFG Ethernet
Frame
IFG

Full-duplex
59
 There are exactly two stations connected with
a full-duplex point-to-point link.
 Both stations on the LAN are capable of, and
have been configured to use, the full-duplex
mode of operation.
 Typically:
 Host-to-Switch
 Switch-to-Switch
 Switch-to-Router
Full-duplex

Learning Switches: Learns Source MAC Address
61MAC Address Table
1 1111
switch
1111
2222
3333
4444
Abbreviated
MAC
addresses
11113333
 Switches are also known as
learning bridges or learning
switches.
 A switch has a source address
table (or MAC Address Table) in
cache (RAM) where it stores a
source MAC address after it learns
about them.
 How does it learn source MAC
addresses?
 Whenever a frame enters a switch,
it will first see if the Source
Address (1111) is in it’s table.
 If it is, it resets the timer
(more in a moment).
 If it is NOT in the table it adds
it, with the port number.Soft Polynomials (I) Pvt. Ltd. Subhash Iyer, Program Head

Destination MAC Address: Filter or Flood
62MAC Address Table
1 1111
switch
1111
2222
3333
4444
Abbreviated
MAC
addresses
11113333
 Next, the switch examines the
source address table for the
Destination MAC address.
 If it finds a match, it filters the
frame by only sending it out that
port.
 If there is not a match if floods it
out all ports.
 In this scenario, the switch will
flood the frame out all other ports,
because the Destination Address
is not in the source address table.

Learning Switches: Learns, Filter or Flood
63MAC Address Table
1 1111 6 3333
switch
1111
2222
3333
4444
Abbreviated
MAC
addresses
33331111
 Most communications involve some
sort of client-server
relationship or exchange of
information. (You will understand
this more as you learn about
TCP/IP.)
 Now 3333 sends data back to 1111.
 The switch sees if it has the Source
Address stored.
 It does NOT so it adds it. (This will
help next time 1111 sends to 3333.)
 Next, it checks the Destination
Address and in our case it can
filter the frame, by sending it only
out port 1.

Destination Address in table, Filter
64MAC Address Table
1 1111 6 3333
switch
1111
2222
3333
4444
Abbreviated
MAC
addresses
11113333
33331111
 Now, because both MAC addresses
are in the switch’s table, any
information exchanged between
1111 and 3333 can be sent
(filtered) out the appropriate port.
 What happens when two
devices send to same
destination?
 What if this was a hub?
 Where is (are) the collision
domain(s) in this example?

 Unlike a hub, a collision does
NOT occur, which would cause the
two PCs to have to retransmit the
frames.
 Collision domains end at the switch
 Instead the switch buffers the
frames and sends them out port #6
one at a time.
 The sending PCs have no idea that
their was another PC wanting to
send to the same destination.
No Collisions in Switch, Buffering
65MAC Address Table
1 1111 6 3333
9 4444
switch
1111
2222
3333
4444
Abbreviated
MAC
addresses
11113333
44443333

MAC Duplex – No collisions
66
 When there is only one device on a
switch port, the collision domain is
only between the PC and the switch,
which is non-existent with full-
duplex.
 With a full-duplex PC and switch
port, there will be no collision, since
the devices and the medium can
send and receive at the same time.
MAC Address Table
1 1111 6 3333
9 4444
switch
1111
2222
3333
4444
Abbreviated
MAC
addresses
11113333
44443333
No Collision Domains

Other Information
67
 How long are addresses kept in the
Source Address Table?
 5 minutes is common on most
vendor switches.
 How do computers know the
Destination MAC address?
 ARP Caches and ARP
Requests (later)
 How many addresses can be kept in
the table?
 Depends on the size of the cache,
but 1,024 addresses is common.
 What about Layer 2 broadcasts?
 Layer 2 broadcasts (DA = all
1’s) is flooded out all ports.
MAC Address Table
1 1111 6 3333
9 4444
switch
1111
2222
3333
4444
Abbreviated
MAC
addresses

What happens here?
68
 Notice the Source
Address Table has
multiple entries for
port #1.
33331111
3333
1111
MAC Address Table
1 1111 6 3333
1 2222 1 5555
2222 5555

What happens here?
69
 The switch resets the 5
minute timer on the
source port entry.
 The switch filters the
frame out port #1.
 But the hub is only a
layer 1 device, so a hub
floods it out all ports.
 Where is the collision
domain?
33331111
3333
1111
MAC Address Table
1 1111 6 3333
1 2222 1 5555
2222 5555
Reset timer
Filter

What happens here?
70
33331111
3333
1111
Source Address Table
1 1111 6 3333
1 2222 1 5555
2222 5555
Collision Domain

Unshielded Twisted Pair (UTP)
72
Straight-through Cross-over Rollover
www.cisco.com/warp/ public/701/14.html

UTP Straight-through Cable
73
 The cable that connects from the switch port to the
computer NIC port is called a straight-through cable.
 Connects unlike devices.
Host or RouterHub or Switch

UTP Straight-through Cable
74
Host or RouterHub or Switch

UTP Cross-over Cable
75
 The cable that connects from one switch port to
another switch port is called a crossover cable.
 Connects like devices.
Hub or Switch Hub or Switch

UTP Cross-over Cable
76

Cabling – Show the straight-through and cross-over cables
77
switch switch
router

Cabling – Show the straight-through and cross-over cables
78
switch switch
router
Straight-through cable
Cross-over cable

Configuring Speed and Duplex
79
 Negotiation between NIC and
switch port.
 Duplex: Full-duplex or Half-
duplex
 Speed: 10/100/1000 Mbps
 Autonegotiation
 Both sides of a link should have
auto-negotiation on, or both sides
should have it off.

When to Use Ethernet 10/100Mb Auto-
Negotiation – From www.cisco.com
80
 Auto-negotiation is an optional function of the IEEE 802.3u Fast
Ethernet standard that enables devices to automatically exchange
information over a link about speed and duplex abilities.
 Auto-negotiation is targeted at ports which are allocated to areas where
transient users or devices connect to a network.
 For example, many companies provide shared offices or cubes for Account
Managers and System Engineers to use when they are in the office rather
than on the road.
 Each office or cube will have an Ethernet port permanently connected to
the office's network.
 Because it may not be possible to ensure that every user has either a 10Mb, a
100Mb Ethernet, or a 10/100Mb card in their laptop, the switch ports that
handle these connections must be able to negotiate their speed and
duplex mode.
 The alternative would be to provide both a 10Mb and a 100Mb port in each
office or cube and label them accordingly.

When to Use Ethernet 10/100Mb Auto-
Negotiation – From www.cisco.com
81
 One of the most common causes of performance issues on
10/100Mb Ethernet links is when one port on the link is operating at half-
duplex while the other port is operating at full-duplex.
 This occasionally happens when one or both ports on a link are reset and
the auto-negotiation process doesn't result in both link partners having
the same configuration.
 It also happens when users reconfigure one side of a link and forget to
reconfigure the other side.
 Both sides of a link should have auto-negotiation on, or both sides should
have it off.
 Our current recommendation is to leave auto-negotiation on for those
devices compliant with 802.3u.
 Many performance-related support calls will be avoided by correctly
configuring auto-negotiation.

Half-duplex, Full-duplex Issue
82
 Switch A, the half-duplex end will sense a neutral carrier and send frames
 Switch B, the full-duplex end, senses the non-neutral carrier and since it doesn’t
care because it is configured as full-duplex, it transmits anyways.
 Switch A senses a collision (the half-duplex side) and stops sending the frame.
 Switch B (the full-duplex side) doesn’t care and keeps on sending frames.
 Data ends up being transmitted only one-way most of the time, with collisions
constantly happening on Switch A, causing performance issues on the network.
(Remember, most network communications is bi-directional.
 This is also a common cause for late collisions (a collision that occurs after the
first 512 bits (slot time) have been sent and the sender believes it has acquired the
channel.
Half-duplex Full-duplex
Switch A or Hub
A
Switch B

Real World Troubleshooting - Symptom
83
switch switch switch switch switch switch
switch switch
router
A
B C D
W
X Y Z
 Hosts connected to switches B, C and D can reach each other and the Internet
with no problems.
 However, hosts on X, Y, and Z can either not access hosts on B, C, and D or
the Internet, or if they can it is extremely slow.
Internet

Lights and indicators
84

Real World Troubleshooting – Diagnostics
85
switch switch
router
A
B C D
W
X Y Z
 You notice that a collision light (or looking at some diagnostic output) on Switch
W, port 1 is always on indicating a very large number of collisions detected on
that port.
Internet
Port 1

Real World Troubleshooting – Problem
86
switch switch
router
A
B C D
W
X Y Z
 The problem is that
 Switch A, Port 8 is in Full-duplex mode
 Switch W, Port 1 is in Half-duplex mode
 Switch A sends whenever it wants to without listening first to see if Switch W is
sending.
Internet
Half
Duplex
Port 1
Full
Duplex
Port 8
I’m half-duplex and I
keep seeing
collisions
I’m full-duplex so I
don’t see any
collisions
X

Real World Troubleshooting – Solution
87
switch switch
router
A
B C D
W
X Y Z
 Configure Switch W, Port 1 to be in full duplex, the same as Switch A, Port A.
Internet
Full Duplex
Transmissions
Full
Duplex
Port 8
Full
Duplex
Port 1

Fast and Gigabit Ethernet
 Fast Ethernet (100Mbps) has technology very similar to
10Mbps Ethernet
 Uses different physical layer encoding (4B5B)
 Many NIC’s are 10/100 capable
 Can be used at either speed
 Gigabit Ethernet (1,000Mbps)
 Compatible with lower speeds
 Uses standard framing and CSMA/CD algorithm
 Distances are severely limited
 Typically used for backbones and inter-router connectivity
 Becoming cost competitive
 How much of this bandwidth is realizable
89

Experiences with Ethernet
 Ethernets work best under light loads
 Utilization over 30% is considered heavy
 Network capacity is wasted by collisions
 Most networks are limited to about 200 hosts
 Specification allows for up to 1024
 Most networks are much shorter
 5 to 10 microsecond RTT
 Transport level flow control helps reduce load (number of
back to back packets)
 Ethernet is inexpensive, fast and easy to administer!
90

Ethernet Problems
 Ethernet’s peak utilization is pretty low (like Aloha)
 Peak throughput worst with
 More hosts
 More collisions needed to identify single sender
 Smaller packet sizes
 More frequent arbitration
 Longer links
 Collisions take longer to observe, more wasted bandwidth
 Efficiency is improved by avoiding these conditions
91

Why did Ethernet Win?
 There are LOTS of LAN protocols
 Price
 Performance
 Availability
 Ease of use
 Scalability
92

T H A N K Y O U
Done – Lets Network to the
Real World NOW
93

Sessions 6,7 Ethernet

More Related Content

What's hot (20)

Similar to Sessions 6,7 Ethernet (20)

Sessions 6,7 Ethernet