By Stephen Bucaro
In 1973 Xerox invented Ethernet to solve the problem of transferring data between
computers. Digital Equipment and Intel collaborated with Xerox in 1973 to publish the
DIX networking standard. In the early 1980s Xerox turned over control of the Standard
to the Institute of Electrical and Electronics Engineers (IEEE). The IEEE sets up
committees to define industry standards. The IEEE 802 committee sets the standards
for networking. The IEEE subcommittee 802.3 sets the standard for Ethernet.
An Ethernet packet contains four main parts: The data, the MAC address of the packets
source, the MAC address of the packets destination and an error checking code. When
creating a packet, the sending computer performs a mathematical operation on the data
and attaches the result, called the cyclic redundancy check (CRC), to the packet. The
receiving computer performs the same operation on the data and compares its result
with the CRC. If the two results do not match, an error has occurred in transmission
of the data. The receiving computer will send a request to the sending computer to
retransmit the packet.
Ethernet networks use the access method known as Carrier Sense Multiple Access with
Collision Detection (CSMA/CD). Multiple access means that several computers share
the same cable. But only one computer can place a data packet on the cable at a given
instant. Carrier sense means that before a computer places a data packet on the
cable, it checks or senses the cable to make sure no other computer is using the
cable. If it senses traffic on the cable, it will wait until the cable is free. When
the computer senses that there is no traffic on the cable, it will send out its data packet.
If two computers sense that there is not traffic on the cable and start transmitting
at exactly the same instant, the data packets will collide. Collision detection
means that both computers can detect if a collision has occurred. If they detect a
collision, both computers stop transmitting. Each computer then generates a random
number and waits for a period of time equal to that number before attempting to
On very long cables, beyond 2,500 meters, computers on opposite ends of the cable
cannot detect the start of a transmission on the opposite end of the cable quick
enough to avoid an undetected collision. An undetected collision will result in
corruption of the data packet.
You might think a system designed with the possibility of having data packet
collisions is somewhat crude. But data collisions occur rarely during normal
operation of an Ethernet network and they have little affect on the efficiency of the
network. As network traffic increases, collisions occur more frequently and have the
effect of limiting the bandwidth (maximum operating speed) of the network.
A modification of the CSMA/CD access method features collision avoidance. In the
Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) access method, a
computer broadcasts a message on the network to signal its intent to transmit before
it places a data packet on the cable. The other computers on the network then refrain
from transmitting data, thus avoiding collisions. However, the broadcasting of
messages increases the amount of traffic on the cable, resulting in little or no
performance gain over CSMA/CD.
More Networking Topologies Articles:
• ISP Multihoming Explained
• Introduction to ATM (Asynchronous Transfer Mode) Networks
• How Do Fiber Optic Couplers Work and How are They Made?
• What Are Fiber Optic Attenuators?
• Transparent Bridging and MAC Address Filtering
• Distance Vector vs. Link State vs. Hybrid Routing
• System Area Network Interface Cards
• The IEEE 802.3 Ethernet Standards
• How to Set up a Private Network
• What Are Fiber Optic Isolators?