Information Technology Reference
In-Depth Information
As with any venture that relies upon sharing, Ethernet networks
require attention to several details in order to make communication
reliable. First, because all computers in an Ethernet network are at
tached to the same cable, any communication from one computer
will be received by all others. Information typically is supposed to
go from one machine to a designated second machine, so the use of
the common medium requires that all communications contain a
header, indicating the intended recipient. With this header, all com
puters will receive a message, but then will look at the header. If the
information is intended for them, they keep the full message; other
wise, they discard the information as they receive it. Thus, while all
computers actually receive all communications, only the intended
computer normally saves material intended for itself.
Second, Ethernet networks require special rules or protocols so
that one computer does not interfere with communications by an
other. For example, when one computer wishes to send a message
to another, it first listens to the cable to determine whether another
machine is already sending a message. If so, the first machine waits
some time before trying again. This rule suggests that once comput
ers start sending a message along the common wire, others will not
interrupt or interfere. Even with this rule of noninterference, how
ever, it still could happen that two machines might want to send
messages at the same time. They both could listen to the cable, de
termine the wire was free, and then start broadcasting their mes
sages. Of course, this would mean that each transmission would be
jumbled by the other. To handle this circumstance, Ethernet proto
cols require each computer to listen to the network as it transmits
its messages—not just before the transmission starts. As it listens to
transmissions on the network, a computer should receive the same
message that it sends on the cable. If the message received is differ
ent, then another computer started sending at the same time. When
two computers simultaneously send messages, each transmitting
computer is required to stop transmitting, wait a random amount of
time, and start the process again. The stopping clears the cable for
later communications. Then, because the two computers wait ran
dom amounts of time before trying again, the time one chooses to
try again will differ from the time the other tries to start. This
should allow both to take their turns with minimal conflict.
Yet another concern regarding sharing could arise if one com
puter tried to dominate the cable—sending a steady stream of infor
