1

Ethernet

``The diagram ... was drawn by Dr. Robert M. Metcalfe in 1976 to presentEthernet ... to the National Computer Conference in June of that year. Onthe drawing are the original terms for describing Ethernet. Since then otherterms have come into usage among Ethernet enthusiasts.''The Ethernet Sourcebook, ed. Robyn E. Shotwell (New York: North-Holland, 1985), title page.

Source for this lecture: http://www.ethermanage.com/ethernet/ethernet.html

Read Forouzan Pages 43-49

2

Invention of Ethernet“In late 1972, Metcalfe and his Xerox PARC colleagues developed thefirst experimental Ethernet system to interconnect the Xerox Alto, apersonal workstation with a graphical user interface. The experimentalEthernet was used to link Altos to one another, and to servers and laserprinters. The signal clock for the experimental Ethernet interface wasderived from the Alto's system clock, which resulted in a datatransmission rate on the experimental Ethernet of 2.94 Mbps.

Metcalfe's first experimental network was called the Alto Aloha Network.In 1973 Metcalfe changed the name to "Ethernet," to make it clear thatthe system could support any computer--not just Altos--and to point outthat his new network mechanisms had evolved well beyond the Alohasystem. He chose to base the name on the word "ether" as a way ofdescribing an essential feature of the system: the physical medium (i.e.,a cable) carries bits to all stations, much the same way that the old"luminiferous ether" was once thought to propagate electromagneticwaves through space. Thus, Ethernet was born.”

3

Ethernet Standards• Formal specifications for Ethernet were published in 1980 by a multi-vendor

consortium that created the DEC-Intel-Xerox (DIX) standard. This effortturned the experimental Ethernet into an open, production-quality Ethernetsystem that operates at 10-Mbps. Ethernet technology was then adopted forstandardization by the LAN standards committee of the Institute of Electricaland Electronics Engineers (IEEE 802).

• The IEEE standard was first published in 1985, with the formal title of "IEEE802.3 Carrier Sense Multiple Access with Collision Detection (CSMA/CD)Access Method and Physical Layer Specifications." The IEEE standard hassince been adopted by the International Organization for Standardization(ISO), which makes it a worldwide networking standard.

• The IEEE standard provides an "Ethernet like" system based on the originalDIX Ethernet technology. All Ethernet equipment since 1985 is builtaccording to the IEEE 802.3 standard, which is pronounced "eight oh twodot three." To be absolutely accurate, then, we should refer to Ethernetequipment as "IEEE 802.3 CSMA/CD" technology. However, most of theworld still knows it by the original name of Ethernet, and that's what we'll callit as well.

4

Ethernet System ElementsThe Ethernet system consists of three basic

elements:1. the physical medium used to carry Ethernet

signals between computers,2. a set of medium access control rules embedded

in each Ethernet interface that allow multiplecomputers to fairly arbitrate access to the sharedEthernet channel, and

3. an Ethernet frame that consists of astandardized set of bits used to carry data overthe system.

5

CSMA/CD Protocol• Access to the shared channel is determined by the medium access control

(MAC) mechanism which is based on a system called Carrier SenseMultiple Access with Collision Detection (CSMA/CD).

• Each interface must wait until there is no signal on the channel, then it canbegin transmitting. If some other interface is transmitting there will be asignal on the channel, which is called carrier. All other interfaces must waituntil carrier ceases before trying to transmit, and this process is calledCarrier Sense.

• No one gets a higher priority than anyone else, and democracy reigns. Thisis what is meant by Multiple Access

• Since signals take a finite time to travel from one end of an Ethernet systemto the other, the first bits of a transmitted frame do not reach all parts of thenetwork simultaneously. Therefore, it's possible for two interfaces to sensethat the network is idle and to start transmitting their frames simultaneously.When this happens, the Ethernet system has a way to sense the "collision"of signals and to stop the transmission and resend the frames. This is calledCollision Detection.

6

Carrier Sense Multiple Access with Collision Detection (CSMA/CD)

Monitor to ensure wire free before transmitted

Monitor during transmission to make sure not garbled by acollision

Once collision detected stop transmission

7

Collision• If more than one station happens to transmit on the Ethernet

channel at the same moment, then the signals are said to collide.The stations are notified of this event, and instantly reschedule theirtransmission using a specially designed backoff algorithm. As part ofthis algorithm the stations involved each choose a random timeinterval to schedule the retransmission of the frame, which keepsthe stations from making transmission attempts in lock step.

• Repeated collisions for a given packet transmission attempt indicatea busy network. The expanding backoff process, formally known as"truncated binary exponential backoff," is a clever feature of theEthernet MAC that provides an automatic method for stations toadjust to traffic conditions on the network. Only after 16 consecutivecollisions for a given transmission attempt will the interface finallydiscard the Ethernet packet.

8

Binary exponential backoff retransmission algorithm

Station encounters first collision: randomly select numberfrom set {0,1}

Other station(s) involved in collision select a random valuefrom a set corresponding to number of collisions this otherstation has experienced.

These randomly chosen values represent how many “slottimes” a station must wait before attempting aretransmission.

An Ethernet slot time is 512 bits (64 bytes). For 10 mbit/sEthernet one bit takes 0.1 microseconds to transmit => slottime is 51.2 microseconds.

Note: Minimum size Ethernet frame is 64 bytes. This includessource and destination MAC, type, and CRC fields

9

Round Size Elements

of set

1 2 {0,1}

2 4 {0,1,2,3}

3 8 {0,1,2,3….,7}

4 16 {0,1,2,3………..,15}

5 32 {0,………………………..,31}

6 64 {0,……………………………….,63}

7 128 {0,……………………………………..,127}

8 256 {0,…………………………………………….,255}

9 512 {0,……………………………………………………,511}

10 1024 {0,…………………………………………………………,1023}

For round 11 through 16 {0,…………………………………………………….,1023}

17 => Excess collision error

The First 10 Rounds of Binary

Exponential Backoff Algorithm

10

Every time a station successfully transmits a frame, it gets to resetits backoff choice set to {0,1}.

half duplex -- only one end station may transmit at the same time.

full duplex – both end stations may transmit at the same time ifdesired.

11

Ethernet Frame Format

Original Metcalfe protocol adopted by Digital Equipment, Intel,Xerox. => “DIX” Ethernet.

Ethernet 2 is updated same DIX frame format.

IEEE Standard 802.3 Not same format

Preamble DA SA T DATA CRC Interframe GAP

46 to 1,500Ethernet: <8> <6> <6> <2> bytes <4> <12>

If FailDiscard

<bytes>

Preamble DA SA L DATA CRC Interframe GAP802.3 <8> <6> <6> <2> <4> <12>

}

**Some more stuff here!

12

Minimum Length• Require a minimum length for CSMA/CD to work

correctly• If complete frame is sent before a collision is

detected by sender, too late for sender to repeatbecause MAC layer has discarded thecompleted frame.

• Smallest frame length allowed by standard is aframe of length 51.2 microseconds which is 51.2microseconds x 10 mbits/seconds = 512 bits (64bytes)

• For 18 bytes of header and the 4 CRC bytes oftrailer, 64-18=46 data bytes minimum

13

Slot Time

• Slot time in Ethernet is defined as theround trip time plus the time to send thejam sequence

• Slot time is defined in bits and is definedas 512 bits

• Slot time in 10 mbit/sec ethernet is 51.2microseconds

14

Jam sequence• Assuming a minimum size ethernet frame of 512

bytes• If a sender at one end sends a frame and

another sender at the far end also sends aframe, need for the senders to be able to see thecollision and to send a jam sequence to informother stations of the collision.

• The round trip time plus the time to send out thejam sequence should be less than the time for asender to send the minimum frame. Sendermust be able to know about the collision before itsends the entire frame.

15

Maximum Network Length• Maximum length = propagation speed x (slot

time / 2)• In wire, electricity travels at 2/3 the rate for travel

in air => 2 x 108 m/s• Maximum length for 10 mbit/sec ethernet is

(2x108) x ((51.2 x 10-6)/2) = 5120 meters• Taking into account the delay needed for

repeaters, jammers, etc the standard uses only2,500 meters as the maximum cable lengthallowed.

16

Maximum length

• The standard defines the maximumallowed frame length to be 1518 bytes.Removing the 18 bytes of header andtrailer, data may be no longer than 1,500bytes.

• Maximum limit was due to 1) memoryexpense so wanted a small buffer and to2) prevent one computer from dominatingthe medium

17

MAC Addresses• The first two fields in the frame carry 48-bit addresses, called the

destination and source addresses. The IEEE controls theassignment of these addresses by administering a portion of theaddress field. The IEEE does this by providing 24-bit identifierscalled "Organizationally Unique Identifiers" (OUIs), since a unique24-bit identifier is assigned to each organization that wishes to buildEthernet interfaces. The organization, in turn, creates 48-bitaddresses using the assigned OUI as the first 24 bits of the address.This 48-bit address is also known as the physical address, hardwareaddress, or MAC address.

• A unique 48-bit address is commonly pre-assigned to each Ethernetinterface when it is manufactured, which vastly simplifies the setupand operation of the network.

18

Multicast and Broadcast• A multicast address allows a single Ethernet frame to be

received by a group of stations. Network software canset a station's Ethernet interface to listen for specificmulticast addresses. This makes it possible for a set ofstations to be assigned to a multicast group which hasbeen given a specific multicast address. A single packetsent to the multicast address assigned to that group willthen be received by all stations in that group.

• There is also the special case of the multicast addressknown as the broadcast address, which is the 48-bitaddress of all ones. All Ethernet interfaces that see aframe with this destination address will read the frame inand deliver it to the networking software on thecomputer.

19

Preamble: AA AA AA AA AA AA AA AB(64 bits)Interfame GAP 96 bits 10 Mbits => 9.6 µs GAP helps improvefairness of collision detection algorithm.

Ethernet T => Type Field Identifies Protocol IE 0x0800 => IP

How can we tell which type (DIX or 802.3) frame we received?Look at 2 bytes after MAC fields in the received Ethernet frame:•If this 2 byte field ≤ 1500 this is a valid length and we have an802.3 frame. Since this is 802.3 we have a “Logical Link Control(LLC) subnetwork attachment point (SNAP) Header” which maybe used for other purposes with other protocols (like SNA).•If this 2 byte field ≥ 1501 This is DIX frame format. No typefields in DIX are ever less than 1501. Thus if you see ≥ 1501, it isa DIX frame. Example 0x0800 is a value ≥ 1501.

20

AA AA 03 000000 DIX Type <1> <1> <1> <3> <2>

802.3 Special Case => Repackage Old DIX Type Field

IE 0x0800

**This is the someother stuff from 2slides previous.

21

Purpose of Ethernet

• A given Ethernet system can carry severaldifferent kinds of high-level protocol data.For example, a single Ethernet can carrydata between computers in the form ofTCP/IP protocols as well as Novell orAppleTalk protocols. The Ethernet issimply a trucking system that carriespackages of data between computers; itdoesn't care what is inside the packages.

22

What is the maximum number of Ethernet frames per second on a 100,000,000 bit per second(One Hundred Megabit per Second) Ethernet network when each Ethernet frame carries 46bytes of IP data?

The size of the Ethernet frame is

6 bytes of Destination MAC Address + 6 Bytes of Source MAC Address + 2 Bytes of Type +46 Bytes of IP data + 4 Bytes of CRC = 64 Byte Ethernet Frames

Each Ethernet frame must have 8 Bytes of Preamble and 12 Bytes of Interframe Gap resultingin 8 + 12 + 64 = 84 bytes transmitted on the wire for each Ethernet frame transmitted.

100,000,000 Bits/sec x (1 Byte/8 Bits) x (1 Frame/84 Bytes) = 148,809.5

From Spirent SmartBits 2000Port-Pair Throughput=====================

Frame size 64 100 200 300 1000 Ethernet Frame Size in Bytes (Includes overhead byts but not Preamble or Interframe Gap

100Mb MaxRate 148810 104167 56818 39063 12255 Theoretical Maximum Possible Number Frames per second

Avg % passed 17.53 25.35 43.80 65.02 99.68 Percent of the Max number possible that actually passed------------------------------------------------------ 5 to 6 26089 26406 24888 25400 12216 Number of frames that passed our Linux Router

23

Trivia: Ethernet transmits the most significant byte first but in eachbyte it transmits the least significant bit first.

Important: Ethernet Cables

Computer 1 HUB Computer 2

Straight-through PatchCables

Computer 1 Computer 1

Need to use a cross overcable.

24

Because one pair of wires used to transmit, another pair used toreceive transmissions. Must swap transmit/receive.

Trivia: In 1979 Metcalfe founded 3 COM Corporation:

computers communicate compatibility

Neighbor Address Discovery

For IP over Ethernet, need a way to discover MAC-Layeraddresses of its IP neighbors.

25

We use an address resolution protocol (ARP) cache. IP uses thishelper protocol ARP that runs directly over Ethernet and doesnot include an IP header.

ARP is in RFC-826

ARP sends a frame to the broadcast address (of all 48 bits one).Included in this is the target’s protocol address (IP address).This is the (IP) address of the end station for which we want theMAC address.

ONLY the target machine sends its MAC address back torequester.

One requester receives ARP reply, it stores it for some hold time.Common time is from 15 minutes to 24 hours. (No standardtime in RFC-826)

26

MAC DA = Broadcast

MAC SA =

my hardware address

ARP Etype ( 0806 )

Enet HW Type (1)

Prot Type ( 0800 ) hLen = 6 pLen = 4

OpCode = 1 (Req) Source’s

Hardware Address

Source’s Protocol Address

Target’s Hardware Address

Target’s Pro-

-tocol Address Padding

Padding

Padding

Padding

Frame Check Sequence (CRC - 32)

FIGURE 8. 3 Ethernet frame carrying ARP packet.

EthermetFrameHeader

ARPPacketPayload

27

C:\>arp -a

Interface: 130.207.231.158 on Interface 2

Internet Address Physical Address Type

130.207.231.1 00-04-de-1b-cf-fc dynamic

130.207.231.12 00-a0-c9-72-52-23 dynamic

130.207.231.161 00-90-27-73-c5-74 dynamic

130.207.231.177 00-a0-c9-31-87-04 dynamic

C:\>

Example ARP table on a Windows PC:

28

LAN interconnection devices operate on “ OSI ” (Open SystemsInterconnection Reference Model) Layers:

Application

Transport

InternetSubnetwork

ApplicationPresentationSessionTransportNetwork

Data Link

Physical

Internet ProtocolStack

Open SystemsInterconnectionReference Model

Layer 1 : Physical LayerLayer 2: Data Link LayerLayer 3: Network Layer

29

OSI Layer 1:

Repeaters ( Not too common now) HUBS ( Very common )

HUB To Backbone Four HUB rule - only 4 hub hops allowed in a “ Collision Domain ”

Hubs are physical layer devices that just “Repeat” what they see, “errored” frames and all.

10 Base-T pinout and Crossover cable HUB attachment Functionality

Endstation1 -- TD+ 2 -- TD- 3 -- RD+ 456 -- RD- 78

Hub Side1 -- TD+2 -- TD-3 -- RD+456 -- RD-78

Endstation1 -- TD+ 2 -- TD- 3 -- RD+ 456 -- RD- 78

Endstation1 TD+2 TD-3 RD+456 RD-78

30

OSI Layer 2:Bridges ( Not common now )Switches ( What we call bridges now )Layer - 2 switches ( Way cool marketing term)

Forwarding decisions based only on data link layer header, that is the MAC Destination Address. Use a table made from observing which addresses are seen on each port.No more than 7 bridges in diameter.Today’s ethernet environments use lots of ethernet switches, reducing the collision domain.Collision domain is the set of ethernet segments that can directly hear each others frames.Worst case is all inputs want to go out on the same output. Must have some buffering and some sort of fairness algorithm inside

31

10 megabit/second

• The original standardized Ethernet system operates at 10-Mbps, and there are fourbase band media segments defined in the 10-Mbps Ethernet standard.

• The four media types are shown with their IEEE shorthand identifiers. The IEEEidentifiers include three pieces of information. The first item, "10", stands for themedia speed of 10-Mbps. The word "BASE" stands for "baseband," which is a type ofsignaling. Baseband signaling simply means that Ethernet signals are the onlysignals carried over the media system.

• The third part of the identifier provides a rough indication of segment type or length.For thick coax the "5" indicates the 500 meter maximum length allowed for individualsegments of thick coaxial cable. For thin coax the "2" is rounded up from the 185meter maximum length for individual thin coaxial segments. The "T' and "F" stand for"twisted-pair" and "fiber optic," and simply indicate the cable type.

• The thick coaxial media segment was the first one defined in the earliest Ethernetspecifications. Next came the thin coaxial segment, followed by the twisted-pair andfiber optic media segments. The twisted-pair segment type is the most widely usedtoday for making network connections to the desktop.

32

100 megabit/second• Compared to the 10-Mbps specifications, the 100-Mbps

system results in a factor of ten reduction in the bit-time,which is the amount of time it takes to transmit a bit onthe Ethernet channel. This produces a tenfold increasein the speed of the packets over the media system.However, the other important aspects of the Ethernetsystem including the frame format, the amount of data aframe may carry, and the media access controlmechanism, are all unchanged.

• The Fast Ethernet specifications include mechanisms forAuto-Negotiation of the media speed. This makes itpossible for vendors to provide dual-speed Ethernetinterfaces that can be installed and run at either 10-Mbpsor 100-Mbps automatically.

33

100 megabit/second

• The first item, "100", stands for the media speed of 100-Mbps. The"BASE" stands for "baseband," which is a type of signaling.Baseband signaling simply means that Ethernet signals are the onlysignals carried over the media system.

• The third part of the identifier provides an indication of the segmenttype. The "T4" segment type is a twisted-pair segment that usesfour pairs of telephone-grade twisted-pair wire. The "TX" segmenttype is a twisted-pair segment that uses two pairs of wires (This iswhat we are using in the lab). The "FX" segment type is a fiber opticlink segment based on the fiber optic physical medium standardthat uses two strands of fiber cable. The TX and FX mediumstandards are collectively known as 100BASE-X.