CSC 805 Assignment
CSC 805 Assignment
Ethernet is the basis of LAN networks. The current LAN market is characterized by an, up to
now, unknown degree of standardization on Ethernet. Due to its huge market share, Ethernet,
despite some disadvantages, scores over all alternative technologies.
Ethernet as we know it today—the industry standard for local-area networking—was born in
May of that year (PBS 1998). Twenty-five years later, Ethernet has been neither dis-proven,
as the theory of ether was by Michelson and Morley, nor discarded, as the PARC research
center was by Xerox. Instead, it is a $10 billion dollar industry that shows no signs of slowing
down even after a quarter of a century—an eternity to most in the computer industry.
Ethernet’s success has been nothing short of phenomenal. For the past three years, Ethernet
and its successor, Fast Ethernet, have retained greater than 85% market share in the network
interface card (NIC) industry—the most visible component of Ethernet hardware (Comm
Serv 1998). The next closest competitor—IBM’s Token Ring—has not been a major threat
for almost a decade. Ethernet usage continues to grow at rates of 20-25% a year, and the most
well-known networking equipment provider, 3Com Corporation, is fast on its way to
becoming the largest, with revenues in excess of $5.4 billion (3Com 1998).
That something so old is still in wide use today, and only now appears to be reaching
technological maturity, strikes many not only as phenomenal, but also as quite anomalous. A
few explanations could be given to explain why computers today are using a technology
invented before the PC itself. First, it could be that Ethernet was not technically or
commercially feasible until recently, and that the concept lay dormant for many years.
However, large distributed networks were using Ethernet technology almost a decade ago;
even before Ethernet was standardized, it was seen as a valid option for local area networks
(LANs). Perhaps the need for LANs themselves only became apparent a few years ago. This
explanation, too, is questionable: early network implementations like ARPANET and
ALOHAnet were being explored, as well as implemented, as far back as the '60s, before
Ethernet was even invented. Maybe Ethernet is only a fad, the latest in a series of Silicon
Valley technologies that come and go. The revolutions in network technology that have
occurred every two to three years, however, suggest otherwise. The network transformation
1
from shared to switched, coaxial to twisted-pair, and megabit to gigabit are rebirths, rather
than replacements, of the pre-existing Ethernet model.
1.1. A short historical overview:
• 1980: Digital Equipment Corporation, Intel and Xerox released the first Ethernet
specification, version 1.0, under the name Ethernet Blue Book or DIX standard. It defines
“Thick Ethernet” in case of 10Mbps CSMA/CD. The first Ethernet controllers, based on the
DIX standard, were available starting from 1982. The second and final version of the DIX
standard, version 2.0, was released in November 1982: Ethernet II.
• 1983: The Institute of Electrical and Electronic Engineers (IEEE) launches the first IEEE
Standard for Ethernet technology. It was developed by the 802.3 group of the IEEE802
Committee and this under the name IEEE802.3 Carrier Sense Multiple Access with Collision
Detection Access Method and Physical Layer Specifications. IEEE reworked some parts of
the DIX standard, especially with regard to the definition of the frame determination.
• 1985: IEEE802.3a; definition of thin Ethernet, cheapernet or 10Base2
• 1987: IEEE802.3d; Fiber Optic Inter Repeater Link (FOIRL). Use of two fibre optic cables
to extend the distance between 10 Mbps repeaters up to 1000m.
• 1987: IEEE802.3e; 1Mbps over twisted pair
• 1990: IEEE802.3i; release of the popular 10Base-T; 10Mbps over UTP category 3
• 1993: IEEE802.3j; 10Base-F: distances greater than 2 km over fiber optic
• 1995: IEEE802.3u; 100Base-T and 100Base-F
• 1997: IEEE802.3x: full-duplex Ethernet
• 1997: IEEE802.3y; 100Base-T2
• 1998: IEEE802.3z; 1000Base-X standard; generally known by the name Gigabit Ethernet
• 1999: IEEE802.3ab; Gigabit Ethernet over twisted pair
• 1999: IEEE802.3ac; 802.1Q: definition of the Q tag with VLAN and priority information.
• 2003: IEEE802.3af; Power over Ethernet
• 2006: IEEE802.3an; 10GBase-T
• 2006: IEEE802.3aq; 10GBase-LRM, Ethernet over multi-mode fiber
Ethernet is only a specification of layers 1 and 2 in the OSI model. It is not a complete
network
Protocol but a sub-net on which other protocols such as the TCP/IP suite can work.
2
- Sending and receiving the serial bit streams over the medium.
- Detecting collisions.
Filling in of the data link layer
- MAC sublayer:
Access mechanism to the network (CSMA/CD).
Building of the data frames.
- LLC sublayer:
Data reliability.
Supply data channels for higher-level applications.
Ethernet is becoming an important carrier network technology. For many years, it was
relegated to office LANs (local area network), connecting PCs, servers, and printers. Recent
high-speed implementations, however, make Ethernet a viable candidate to provide new
carrier-based services such as:
Voice over IP (Vo0IP), a technology that enables voice calls over data networks. This
may one day eliminate the need for separate voice and data facilities.
Metropolitan area networks (MANs), high-bandwidth pipes that can link company
data
centres over a 15 to 20 mile area.
Ethernet in the First Mile (EFM), an emerging standard that may compete with DSL
and cable modems to bring voice, video, and data to homes.
3
New topologies: Cabling for the newer Ethernet standards uses a star, a bus-star
hybrid called a tree, and even a ring.
Increased bandwidth: The standards now define speeds between 1 Mbps and 1 Gbps
(soon 10 Gbps).
Support for full-duplex operation: The original standard supported half-duplex only.
(On full-duplex networks, CSMA/CD is not required.)
Expansion of the distances supported: Ethernet is no longer restricted to the LAN. It is
now deployed in MAN networks, and will soon provide the underlying service in
WAN (wide area network) environments as well.
Support for new applications: Gigabit and 10 Gigabit Ethernet are able to provide
transport for voice and video as well as data.
Over the years, Ethernet has gained wide acceptance because it offers clear advantages over
competing technologies. Ethernet is:
4
2.1. ETHERNET TOPOLOGY
Linear Bus
The original Ethernet standard specified a linear bus. This topology is seldom used in
new installations. A cable break on a linear bus brings down the whole network, and
cabling costs can be reduced by using twisted pair cables in a star configuration.
Star
The star topology is the most common. It mitigates Ethernet distance limitations, can use
inexpensive unshielded twisted pair cables, and the entire network doesn’t go down if a cable
breaks or is disconnected.
5
Ring
The ring is used in Metropolitan Area Networks to deliver Ethernet using Add/Drop
Multiplexers (ADMs) at customer sites.
The ADMs connect to the LAN router to deliver Ethernet.
6
automatically adjust to the speed used on the network. Many NICs support both half- and
full-duplex operation.
Repeaters and Hubs
Hubs are repeaters that connect two or more Ethernet segments by regenerating the electrical
signal and broadcasting it out all ports. This means that every connected segment is in the
same collision domain. In other words, when one device is transmitting, no other device can
transmit, or collisions will occur. This is in contrast to Ethernet bridges and switches, which
are more discriminating about where they send the transmission.
Ethernet hub
Bridges
Like repeaters, bridges straddle two Ethernet segments. Unlike repeaters, they make
intelligent decisions about which frames to forward and which to discard. Bridges reduce
LAN traffic by dividing it into two segments. They perform a service similar to switches,
though most often bridges support one network boundary only; switches support four or more
segments.
Switches
Though they are multi-port devices like hubs, switches are multiport bridges. Rather than
broadcasting a frame out every port as hubs do, they forward the frame to its intended
destination only. This means that each port becomes a separate collision domain. Bandwidth
is shared only with stations using that port. Ports that host only a single station can be
7
configured for full-duplex communication, which means collisions can’t occur. This
arrangement also means that bandwidth doubles:
Routers
The task of defining a Local Area Network (LAN) domain is accomplished using a router.
Routers are located at the service provider’s central office and interface with the LAN router
located at the customer’s premises. Routers pass traffic only to the intended destinations, and
block all broadcasts as configured. Multiple routers are common within the customer’s LAN
domain, used as needed to segment large LAN installations. The Internet is built using many
thousands of routers that define all networks and services that make up this vast global
information resource.
8
Line Coding
10 Mbps, 100 Mbps, and 1 Gbps Ethernet technologies utilize a specific type of line coding
referred to as 8B/10B. 8B/10B conversion is a simple algorithm that converts 8 bits into 10
bits, producing a 25% overhead. This effectively increases the line rate without increasing the
actual throughput. There are two main reasons for the use of this particular type of line
coding.
DC Current Elimination: In order to keep the average DC current on the wire
equal to zero, there must be an equal number of positive pulses and negative
pulses. The algorithm for 8B/10B converts the 8-bit sequence, where a positive or
negative DC voltage would present itself. Those 8 bits are converted to a 10-bit
sequence, eliminating the DC current artifact.
Clocking: There is no common clock within an Ethernet circuit. The bit stream
from one device drives the clock on the receiving device. If there are too many
zeros (absence of pulses) in the bit stream, the receiving device will lose
synchronization. The 8B/10B conversion takes the 8 bits without enough clocking
pulses and converts it to 10 bits without enough positive pulses.
In terms of testing, there are no options or settings for line coding. All test sets for Ethernet
support this type of line coding.
Duplex Options
Another important physical layer characteristic of Ethernet is its full-duplex or half-duplex
operation. A full-duplex circuit is able to transmit and receive at the same time, similar to a
telephone where a person can speak and hear at the same time. A half-duplex circuit is either
speaking or listening; it is incapable of both operations simultaneously. Since it can only
operate in one direction, a half-duplex circuit only offers the user about half of its actual
bandwidth. In other words, a 100 Mbps circuit running in a half-duplex environment offers
the user only about 50 Mbps worth of actual data throughput.
10 Megabit Ethernet (10 BASE-X) is most commonly deployed in a half-duplex
environment. Users requiring lower bandwidth applications are able to utilize 10 BASE-X
and save the cost of higher bandwidth equipment.
100 Megabit Ethernet (100 BASE-X) can be deployed in either a full-duplex or half-duplex
environment, depending on bandwidth requirements. Most switches deployed in a LAN today
are capable of 100 BASE-X.
9
Gigabit Ethernet (1000 BASE-X) is almost always deployed in a full-duplex environment,
allowing network routers and switches to take full advantage of the bandwidth. The IEEE
802.3 Ethernet standard, however, does allow for 1000 BASE-X deployment in a half-duplex
environment.
10 GigE LAN has a line coding of 64B/66B instead of the 8B/10B coding used for lower
speed Ethernet services. The main reason for the switch from 8B/10B is the bit rate. At 10
Gigabit, an 8B/10B formatted signal has a line rate of 12.5 Gbps. The 64B/66B signal, with
significantly less overhead (3% instead of 25%), has an effective line rate of 10.3 Gbps.
The 10 GigE WAN specification utilizes SONET physical layer standards. Most often, an
NRZ optical pulse is used. There is also a common clock; there-fore, no 64B/66B conversion
is required. The most critical physical layer com-ponent of a 10 GigE WAN signal is the
SONET framing structure. In a 10 GigE WAN signal, the Ethernet packet is placed into a
SONET frame. This frame also includes the section, line, and path overhead. In this type of
structure, there are several overhead bytes that are slightly different than the traditional
specification.
Duplex Options
10 GigE Ethernet (10G BASE-X) must be deployed as a full duplex service. There are no
specifications in the standard allowing for a half-duplex option.
WAN Interface Sublayer (WIS)
One of the main differences between 10 GigE WAN and 10 GigE LAN is the SONET
portion. The WIS is responsible for mapping Ethernet frames into a SONET frame and
demapping Ethernet frames from a SONET frame. Generally, the WIS is designed to:
10
Map Ethernet MAC frames into a SONET frame.
Implement framing, scrambling, and defect/ anomaly (error/alarm) detection to
allow for compatibility with the SONET requirements
Provide a 9.95328 Gbps effective data rate at the service interface, conforming to
the requirements of a SONET STS-192c signal.
Destination and Source Address Fields: The destination and source address fields, as
their names suggest, are the fields in the data frame that identify the destination and
source MAC addresses for the frame. The source address is the device that transmitted
the frame, and the destination address is the device destined to receive the frame.
Frame Type: This field contains information that determines the format of the frame,
either an Ether-type field for Ethernet, Version 2 or a length field for IEEE 802.3.
Data Field: This field contains the bulk of the frame. This is where the upper layer
information is encapsulated.
11
FCS Field: This is the frame check sequence. The FCS is a calculation performed by
the equipment generating the frame on the total bits in the frame. If any of the bits
change while the packet traverses the network, the FCS value will no longer be valid.
The device receiving the frame at the far end will see that the frame has been
corrupted during transmission, and it will discard the frame.
12
reflections of the signals, resulting in a poor transmission quality. This type of
implementation has been superseded. The thick, rigid yellow coax was rapidly replaced by
the black, more flexible coax which resulted in the implementation of thin Ethernet
(10Base2). The connection of the different stations is realized by T-shaped BNC connectors
whereby a maximum segment length of about 200 metres can be applied. Important cabling
detail that is required for most bus technologies: the terminating resistance (terminator) - a
small, cheap device that has to be mounted on all endings of the coax cables that form an
Ethernet. A terminating resistance consists of a resistance that connects the central core of the
cable with the shielding: when an electrical signal reaches the terminating resistance, this is
discarded. For the correct functioning of a network, the terminating resistance is
indispensable as the end of the non-terminated cable reflects electrical signals just as a mirror
reflects light. When a station tries to send a signal over a non-terminated cable, then this
signal will be reflected by the cable end. When the reflection reaches the sending station,
interference will occur.
The MAU, developed for twisted pair, is equipped with 4 data pins: 2 for sending, 2 for
receiving. This is the basis for full duplex Ethernet. In principle, any point to point
13
communication is possible. Therefore, every host has to be connected directly with a
structure element: a hub or a switch.
Fast Ethernet
The UTP cable, e.g. CAT5 (Class 5) UTP (Unshielded Twisted Pair), supports speeds up to
100Mbps. The cable consists of 8 wires, arranged in 4 pairs. The 4 pairs can be identified as 1
is always completely coloured and the other one has the same colour with white parts in
between. Only 2 of the 4 pairs are used in 10/100Base-T (pair 2: orange/white and orange and
pair 3: green/white and green).
The IEEE specification for Ethernet 10/100Base-T requires that the one used pair is
connected to pin 1 and pin 2 of the connector while the second pair is connected to pin 3 and
pin 6. The other two unused pairs will be connected to pin 4 and 5 and on pin 7 and 8.
Table above shows the pin configuration for 10/100Base-T. TD stands for Transmitted Data,
RD stands for Received Data. The plus- and the minus signs indicate that the signal is sent
mirrored via two data lines also see figure below.
14
Transmission technology for 10/100Base-T
The straight-through cable, also called patch cable, is the cable that we get when we connect
both sides of the cable pair 2 with pin 1 and pin 2, while pair 3 is connected with pin 3 and
pin 6. This cable can be used for connections between the patch panel and the hub/switch, the
PC and the hub/switch or the PC and the wall. This cable is generally used for the connection
of a structure element and an end element. A cross-over cable is required to set up the PC-PC
connections (connection of two end elements) and to secure connections between hub/switch
and another hub/switch (connection between two structure elements). In order to make a
cross-over cable, we have to switch the used pairs. Along one side, pair 2 has to be connected
with pin 3 and pin 6 while pair 3 has to be connected with pin 1 and pin 2.
Current Ethernet ports support auto-crossing. This means that it can be detected automatically
which cable is used and the crossing will be corrected internally if necessary.
The IEEE Fast Ethernet has defined 100Base-T as extension on the 10Base-T. Fast Ethernet
is characterised by:
Data transmission at a speed of 100Mbps
Full Duplex communication
Wireless Ethernet
In Fast Ethernet, a mechanism is provided for auto negotiation: this makes it possible to build
Ethernet interfaces that switch automatically between 10Mbps and 100Mbps. For the
10BaseT standard, every data bit is coded in one physical bit. In other words, for a group of
eight data bits, eight signals are generated in the cable. The 10Mbps data rate means a clock
rate of 10MHz. For every clock pulse, one single bit is sent. 100Base-T uses the so-called
15
4B/5B scheme whereby each group of four bits is coded in a 5 bit signal. So, one single bit is
not exactly converted into one single signal in the cable.
Gigabit Ethernet
Gigabit Ethernet targets a data rate of 1000Mbps. If the CAT5 Ethernet cables have to be
used for this, for example, then this causes a problem as they only support a clock rate up to
125MHz. In order to realise this, the technology has to be adapted. First, 1000Base-T codes
two bits per clock signal (00, 01, 10 and 11) and uses four voltage levels for this.
Furthermore, 1000Base-T uses all four data pairs of an Ethernet cable. The four data pairs are
applied here bi-directionally. Data are sent or received via all four data pairs. Gigabit
Ethernet therefore still uses the 100Base-T/Cat5 clock rate of 125MHz. A data rate of
1000Mbps is reached as 2 bits are being processed for every clock pulse and this is done via
four data pairs. This modulation technology is called 4D-PAM5 and currently uses five
different voltage levels. The fifth voltage level is used for the error mechanism. Table2.2
shows the Gigabit Ethernet pin configuration. BI stands for bi-directional while DA, DB, DC
and DD stands for data A, data B, data C and data D.
16
3.1.4. Implementations based on fibre
In order to make longer segment distances possible, the glass fibre cable was integrated as a
suitable interface. The first fibre glass variants are known by the name 10Base-F and
100BaseF. Separate fibres are used all the time for the sending and receiving of data.
Gigabit Ethernet over fibre has been developed for the full-duplex mode with a data rate of
1000Mbps. There are two different variants for Gigabit Ethernet. 1000Base-SX and
1000BaseLX.
1000Base-SX uses light pulses with short wavelength over multimode fibre. 1000Base-LX
uses light pulses with long wavelength over multimode or single-mode fibre.
Recently, 10Gigabit Ethernet over fibre with different variants also has been added.
REFERENCES
Bell, Gordon. “Why Digital is Committed to Ethernet for the Fifth Generation”, a summary
of remarks made at the Xerox/Intel/ Digital Seminar, February 10, 1982.
Report provided by Gordon Bell.
Bell, Gordon, Former Vice President of Engineering, DEC, email interview conducted
by Eden Miller, December 4, 1998, concerning his role promoting Ethernet’s
acceptance at DEC.
Boggs, David R., Mogul, Jeffrey C., & Kent, Christopher A. “Measured Capacity of an
Ethernet: Myths and Reality.” WRL Research Report 88/4 (1988): 1-31.
Boggs, David R., Former engineer at PARC, Xerox, telephone interview conducted
by Peter Sun, December 2, 1998, concering his role in the invention of Ethernet.
Davis, Dwight B. “Ethernet spec spawns diverse group of interface products,” Mini-
Micro Systems, XV, No. 5 (May 1982), 143-151.
Davis, Dwight B.. “Xerox moves to bolster embattled Ethernet,” Mini-Micro Systems,
XV, No. 2 (February 1982), 17-18.
“DEC, Intel, Xerox Form Local Data Com Net Team,” Electronic News, 26 (May 19,
1980), 1.
17
“Desks of America Unite,” Advertisement for Xerox 860 Information Processing
System, Computer Decisions, 14, No.1 (January 1982), 20.
“In Your Future: Local Computer Networks,” EDP Analyzer, 18, No. 6 (June 1980), 1-
13.
Kirsner, Scott. “The Legend of Bob Metcalfe,” Wired, 6.11 (November 1998), 182-
186+.
Kuhn, Thomas. The Structure of Scientifific Revolutions, Chicago: The University of Chi
cago Press, 1962.
Lettieri, Larry. “Joint Effort turns out Ethernet chips, boards,” Mini-Micro Systems, XV,
No. 11 (November 1982), 71-74.
“MDS Operators Rap Xerox Net Proposal,” Electronic News, 25 (January 15, 1979), 25.
Metcalfe, R.M. and D. R. Boggs, “Ethernet: Distributed Packet Switching for Local
Computer Networks,” Xerox PARC Blue and White Series Reports, P7600002,
November 1975.
Metcalfe, Robert M., “How Ethernet Was Invented”, IEEE Annals of the History of
Computing, 16, No. 4, (1994), 81-88.
Metcalfe, Robert M., Former research scientist at PARC, Xerox, telephone interview
conducted by Peter Sun, December 8, 1998, concerning his role in the invention of
Ethernet.
Metcalfe, Rovert M., David R. Boggs, Charles P. Thacker, and Butler W. Lampson.
United States Patent 4,063,220: Multipoint Data Communication System with Col-
lision Detection. December 13, 1977.
“Networking standards. Nobody keeps your options open like Digital,” Advertisement
for DEC, Data Communications, 12 (March 1983), 82.
Norberg, Arthur L. and Judy E. O’Neill. Transforming Computer Technology: Informa-
tion Processing for the Pentagon, 1962-1986. Baltimore: The Johns Hopkins Uni-
versity Press, 1996.
18
Public Broadcasting Service. Nerds 2.0.1. http://www.pbs.org/opb/nerds2.0.1/
serving_suits/3com.html.
Saxton, W.A. and Morris Edwards. “The integrated network movement,” Infosystems,
26, No. 3 (March 1979), 78-80.
Shotwell, Robyn, ed., The Ethernet Sourcebook. New York: Elsevier Science Publish-
ing Company, Inc., 1985.
“The Ethernet Local Network: Three Reports,” Xerox PARC Blue and White Series,
P8000013, February 1980.
19