Information Technology Reference
In-Depth Information
but much higher when employees are working at their desks.
Moreover, some computing activity requires rather little message
communication, whereas other work generates considerable traffic.
For example, word processing largely depends on people typing and
reading, so computers perform little extensive processing, and com
munication is limited to the sending and receiving of small text files.
On the other hand, if people are reading graphical images from the
World Wide Web, much data may need to be communicated over a
network.
All networks typically experience regular variations in the de
gree to which they are used, so the people developing network sys
tems can plan for these changes in demand. They create networks
that are targeted to work their best when responding to only a frac
tion of the demands they are capable of handling. If a network has
sufficient capacity, then it can handle short periods of high traffic.
Otherwise, communication may be slow or even break down during
spikes of high demand.
The difficulties regarding demand are compounded in Ethernet
networks in which all machines use the same cable. For example, as
traffic increases, two machines are more likely to try transmitting at
the same time. When this occurs, they each stop, wait some time,
and try again. The cable is in use for each unsuccessful attempt as
well as for the time used for successful transmission, so when a com
puter has to send a message repeatedly, some potential capacity is
wasted. Experience suggests that if an Ethernet cable is used to more
than 50% to 60% of capacity, noticeable delays may arise, and the
network may become bogged down.
To keep demand relatively low on networks, many organiza
tions maintain separate, small networks, and then provide connec
tions among those networks. This approach forms the basis for
Figure 8.7, and the discussion of that figure illustrates how the ap
proach might apply to star networks.
For Ethernet networks, electrical properties of cables restrict ca
bling between a computer and a hub to 100 meters (about 328 feet).
For greater distances, hubs can be connected together, or devices
called
repeaters
can be used to amplify the electrical signals. Within
a small physical area, computers can connect to a hub, and then
hubs may in turn connect to a more centralized hub, as shown in
Figure 8.7.
