The word network has several definitions. The most commonly used meaning
describes the methods people use to maintain relationships with friends and
business contacts. Applied to computers, the term has a similar definition. A
network is a way to connect computers together so that the' communicate,
exchange information, and pool resources. In business, networks have
revolutionized the use of computer technology. Many businesses that used to rely
on a centralized system with a mainframe and a collection of terminals
(input/output devices that are connect mainframes and do not have the same
features as PCs) now use eom networks in which every employee who needs a
computer has a per computer connected to the network. Computer technology and
expertise no longer centralized in a company's mainframe and information systems
departments. The technology and expertise are distributed throughout
organization among a network of computers and computer literate. In education,
schools have also shifted to strategies built around networks personal
computers. These include LANs (local area networks), and network that connects
the computers and printers in a computer to a WANs (wide area
networks)--especially the Internet.
Whatever the setting, networks provide tremendous benefits. Four most
compelling benefits are:
-
Allowing simultaneous access to critical programs and data
-
Allowing people to share peripheral devices, such as printers and scanners
-
Streamlining personal communication with e-mail
-
Making the backup process easier
The following sections examine each of these advantages in more detail.
Simultaneous Access
It is a fact of business computing that multiple employees, using a computer
network, often need access to the same data at the same time. Without a net-work
that enables file sharing, employees keep separate copies of data on dif-ferent
hard disks, and universally updating the data becomes very difficult. As soon as
a change is made to the data on one machine, a discrepancy arises, and it
quickly becomes difficult to know which set of data is correct. Storing data
that is used by more than one person on a shared storage device makes it
possible to solve the problem.
It is also true that most office workers use the same programs. One solution
to purchasing separate copies of applications for every worker is to use network
versions of programs. These programs are designed so that only one copy of the
application needs to be stored on the network server (called an application
server), with a minimum number of supporting files copied to each employee's
computer. A network server is a large central computer. (If the server stores
data files for users to access, it is commonly called a file server.) When
employees need to use a program, they simply load it from a shared storage
device into the RAM of their own desktop computers, as shown in Figure 7.1.
A network version of a software application is also a more efficient use of
hard disk space because many users can access a single shared copy instead of
storing separate copies on each user's hard disk.
Some software designed for networks is classified as groupware. This type of
software includes scheduling software, e-mail, and document management software.
Groupware allows multiple users on a network to cooperate on projects. Users can
work on the same documents, share their insights, and keep each other abreast of
their schedules so that meetings can be set up easily. Lotus Notes and Microsoft
Exchange are perhaps the best known examples of groupware, although there are
many competitors.
Shared Peripheral Devices
Perhaps the best incentive for small businesses to link com-puters in a
network is to share peripheral devices, especially expensive ones such as laser
printers, large hard disks, and scanners, as shown in Figure 7.2.
Many high quality laser printers cost more than $2,000, so it is not very
cost-effective for each user to have one. Sharing a laser printer on a network
makes the cost much less prohibitive. By using a process called spooling,
multiple users can send multiple print jobs to a printer. (Spooling can also
occur when a computer is not connected to a network, and mul-tiple print jobs
are sent to a non-networked printer.) When users print a document or other file
to a net-worked printer (known as a print job), each job is stored in a
temporary spool file on the file server. As the printer finishes printing a
current job, the file server sends the next spooled job to the printer so that
it can be printed. Typically, a banner page is printed at the beginning of a new
job to separate print jobs.
Personal Communication
One of the most far reaching applications of data communications is electronic
mail (e-mail), a system for exchanging written messages (and increasingly, voice
and video messages) through a network. E-mail is some-thing of a cross between
the postal system and a telephone answering system. In an e-mail system, each
user has a unique address. To send someone an e-mail message, you enter the
person's e-mail address and then type the message. When you are finished, the
message is sent to the e-mail address. The next time that user accesses the
e-mail system, it reports that mail has arrived. Some sys-tems notify the
recipient as each message arrives by flashing a message on the computer screen
or beeping. After reading the message, the recipient can save it, delete it,
forward it to someone else, or respond by sending back a reply mes-sage. Figure
7.3 shows the process for sending and receiving e-mail.
In addition to sending a page or pages of mail text, many systems allow you
to attach data files--such as spreadsheet files or word processed documents--to
your message. This means that an e-mail system allows people to share files even
when they do not have access to the same storage devices. For example, a local
area network also may have a connection to a large information network, such as
America Online, Microsoft Network, or the Internet. In this case, the person on
the local network can share files with anyone on the large information network.
E-mail is both efficient and inexpensive. Users can send written messages
without worrying about whether the other user's computer is cur-rently running.
On centralized networks, the message is delivered almost instantaneously, and
the cost of sending the message is negligible. E-mail has provided the modem
world with an entirely new and immensely valuable form of communication
In addition to e-mail, the spread of networking technology is adding to the
popularity of teleconferencing and videoconferencing. A teleconference is a
virtual meeting in which a group of people in different locations conducts
discussions by typing messages to each other. Each message can be seen by all
the other people in the teleconference. Teleconference software has become more
sophisticated, gradually adding such features as a shared scratch pad where
diagrams or pictures can be drawn or electronically pasted.
The spread of networking is adding to the popularity of collaborative
software, which allows users to connect with one another over LAN or modem links
so that they can see what's happening on other users' computers. It lets people
send messages, exchange files, and sometimes even work on the same document at
the same time. If users have the necessary hardware and software, they can
actually see and speak to each other as they meet online (instead of merely
typing messages). This is a process known as videoconferencing, as shown in
Figure 7.5.
Easier Backup
In business, data is extremely valuable, so making sure that employees
back up their data is critical. One way to address this problem is to keep all
valuable data on a shared storage device that employees access through a
network. Often the person managing the network has the responsibil-ity of making
regular backups of the data on the shared storage device from a single, central
loca-tion. Network backup software is also available that enables backups to be
made of files stored on employees' hard drives. This way, the files do not have
to be copied to the central server to be backed up.
HOW NETWORKS ARE STRUCTURED
To understand the different types of networks and how they operate, it
important to know something about how networks can be structured. Fir there are
two main types of networks, distinguished mainly by geography local area
networks (LANs) and wide area networks (WANs). Second, any o these types can be
classified according to the logical relationships among the computers. There are
networks that use servers (such as file servers an, application servers) and
those that do not (called peer-to-peer networks).
Local Area Networks
A network of computers located relatively near each other and connected
b3 a cable (or a small radio transmitter) is a local area network (LAN). A LAN
can consist of just two or three PCs connected together to share resources, or
it can include several hundred computers of different kinds. Any network that
exists within a single building, or even a group of adjacent buildings, is con-sidered
a LAN.
A LAN permits all the comput-ers connected to it to share hardware, software,
and data. The most commonly shared resources are disk storage devices and
printers. To LAN users, the network is (or should be) completely transparent,
which means that the shared devices on it seem to be directly connected to the
user's com-puter as if they were merely peripherals. For example, a file server
should appear to the LAN user simply as another disk drive.
In addition to shared hardware, LANs can provide all the other benefits of
networks, including simultaneous access, enhanced personal communica-tion, and
easier backup.
Connecting Networks
It is often helpful to connect different LANs. For example, two different
departments in a large business may each have its own LAN, but if there is
enough need for data communication between the departments, then it may be
necessary to create a link between the two LANs.
To understand how this can be accom-plished, you must first know that, on a
network, data is sent in small groups called packets. A packet is a group
of bits that includes a header, payload, and control elements that
are transmitted together (see Figure 7.7). You can think of a packet as one
sentence or a group of numbers being sent at the same time. The payload is the
part that contains the actual data being sent. The header contains information
about the type of data in the payload, the source and des-tination of the data,
and a sequence number so that data from multiple packets can be reassembled at
the receiving computer in the proper order.
Each LAN is governed by a protocol, which is a set of rules and formats for
sending and receiving data. If two LANs are built around the same communication
rules, then they can be connected with a bridge or a router. A
bridge is a relatively simple device that looks at the information in each
packet header and rebroadcasts data that is traveling from one LAN to
another. A router is a more complicated device that stores the
addressing information of each computer on each LAN and uses this
information to act like an electronic post office, sorting data and sending it
along the most expedient route to its destination. Bridges forward data from one
network to another but are not suitable in many large organizations because of
the amount of packets bridges send to all networks connected to the bridge.
Routers, on the other hand, send packets only to the desired network, reducing
overall net-work traffic.
In some cases, a router can also be used to connect two different types of
LANs. However, a router only "routes" data: it knows how to send it to the
correct location, and that's all. If you need a more sophisticated connection
between networks, you need a gateway, a computer system that connects two
networks and translates information from one to the other. Packets from
different networks have different kinds of information in their headers, and the
information can be in different formats. The gateway can take a packet from one
type of network, read the header, and then encapsulate the entire packet into a
new one, adding a header that is understood by the second network, as shown in
Figure 7.8.
Wide Area Networks
Typically, a wide area network (WAN) is two or more LANs connected together,
generally across a wide geographical area using high speed or dedicated
telephone lines. For example, a company may have its corporate headquarters and
manufacturing facility in one city and its marketing office in another. Each
site needs resources, data, and programs locally, but it also needs to share
data with the other site. To accomplish this feat of data communication, the
company can attach a router to each LAN to create a WAN. Figure 7.9 shows a
typical WAN connecting two LANs.
The Internet is the ultimate WAN because it connects many thousands of
computers and LANs around the world. Most of the commercial online ser-vices and
large bulletin boards were not WANs when they started out because, typically,
users dialed in to a single computer or a group of computers housed at a single
site. (Commercial online services are businesses you connect to via modem that
offer their own internal products and services, one of which may be Internet
access.) However, today most of these systems provide connections to other
specialized services and to the Internet, so they are now more like WANs.
File Server Networks
Describing a network as a LAN or a WAN gives a sense of the physical area
the network covers. However, this classification does not tell you anything
about how individual computers on a network, called nodes, interact with other
computers on the network.
Many networks include not only nodes but also a central computer with a large
hard disk that is used for shared storage. This computer is known as the file
server, network server, or simply, server. Files and programs used by more than
one user (at different nodes) are generally kept on the server.
One relatively simple implementation of a network with nodes and a file
server is a file server network. This is a hierarchical arrangement in which
each node can have access to the files on the server but not necessarily to
files on other nodes. When a node needs information on the server, it requests
the entire file containing the information. In other words, the file server is
used simply to store and forward (send) files (see Figure 7.10).
Client/Server Networks
One popular type of server based network is client/server computing, a
hierarchical strategy in which individual computers share the processing and
storage workload with a central server. This type of arrangement requires
specialized software for both the individual node and the network server. It
does not, however, require any specific type of network. Client/ server software
can be used on LANs or WANs and a single client/server pro-gram can be used on a
LAN where all the other software is based on a sim-ple file server relationship.
The most common example of client/server computing involves a database that
can be accessed by many different computers on the network. The data-base is
stored on the network server. Also stored on the server is the server portion of
the database management system (DBMS), the program that allows users to add
information to, or extract it from, the database. The user's computer (which can
be called the node, workstation, or client) stores and runs the client portion
of the DBMS.
Now, suppose that the user wants information from the database. For example,
suppose that the database is a list of customer purchases, and the user needs to
know the names of customers in the Wichita area who made purchases of more than
$500. The user uses the client software to describe the information that is
needed and sends the request to the server. The server soft-ware searches the
database, collects the relevant customer names, and sends them back to the
client. The client software then presents the information to the user in a way
that makes sense. This process is shown in Figure 7.11.
Client/server software is valuable to large, modern organizations because it
distributes processing and storage workloads among resources efficiently. This
means that users get the information they need faster.
Client/server computing is also a commonly used model on the Internet. Users
typically have client software that provides an easily used interface for
interacting with this giant WAN. Other types of processing, such as receiving,
storing, and sending e-mail messages, are carried out by remote computers
running the server part of the relevant software.
Peer-to-Peer Computing
A third arrangement is a peer-to-peer network, in which all nodes on the
network have equal relationships to all others, and all have similar types of
software. Typically, each node has access to at least some of the resources on
all other nodes, so the relationship is nonhierarchical. If they are set up
cor-rectly, Windows 95 and its predecessor, Windows for Workgroups, give users
access to the hard disks and printers attached to other computers in the
net-work. A peer-to-peer network is shown in Figure 7.12.
In addition, some high-end peer-to-peer networks allow distributed com-puting,
which enables users to draw on the processing power of other com-puters in the
network. That means people can transfer tasks that take a lot of CPU power--such
as creating computer software--to available computers, leaving their own
machines free for other work.
Peer-to-peer LANs are commonly set up in small organizations (fewer than 50
employees) or in schools, where the primary benefit of a network is shared
storage, printers, and enhanced communication. Where large data-bases are used,
LANs are more likely to include client/server relationships.
A peer-to-peer network can also include a network server. In this case, a
peer-to-peer LAN is similar to a file server network. The only difference
between them is that the peer-to-peer network gives users greater access to the
other nodes than a file server network does (see Figure 7.12).
NETWORK TOPOLOGIES FOR LANS
In addition to the size of a network and the relationship between the
nodes and server, another distinguishing feature among LANs is the topology--the
of the cables that connect the nodes of the network. There are three basic
topologies: bus, star, and ring. Network designers consider a number of factors
in determining which topology, or combination of topologies, to Among the
factors considered are the type of computers currently the type of cabling
currently in place (if any), the cost of the components and services required to
implement the network, the distance between and the speed with which data must
travel around the network.
Bus Topology
A bus network, like the bus of a computer itself, is a single conduit to
which I the network nodes and peripheral devices are attached (see Figure 7.13).
Nodes on one type of bus network, Ethernet, transmit data at any time, of any
data being sent by other nodes. If one set of data happens to collide with
another set of data transmitted by other nodes--that is, if two nodes try to
send data at the same time--each node waits a small, random amount of time and
then attempts to retransmit the data.
Although the bus topol-ogy is one of the most common, it has inherent
disadvantages. Keeping data transmissions from colliding requires extra
circuitry and software, and a broken connec-tion can bring down (or "crash") all
or part of the network, rendering it inoper-able so that users cannot share data
and peripherals until the con-nection is repaired.
The Star Topology
A star network places a hub in the center of the network nodes. Groups of
data are routed through the central hub to their destinations. This scheme has
an advantage in that the hub monitors traffic and prevents collisions, and a
broken connection does not affect the rest of the network. If you lose the hub,
however, the entire network goes down. Figure 7.14 shows the star topology.
The Ring Topology
The ring topology connects the nodes of the network in a circular chain in
which each node is connected to the next. The final node in the chain con-nects
to the first to complete the ring, as shown in Figure 7.15. With this
methodology, each node examines data that is sent through the ring. If the data
is not addressed to the node examining it, that node passes it along to the next
node in the ring. The ring topology has a substantial advantage over the bus
topology. There's no danger of collisions because data always flows in one
direction. One drawback to the ring, however, is that if a connection is broken,
the entire network goes down.
NETWORK MEDIA AND HARDWARE
No matter what their structure, all networks rely on media to link their
nodes and/or servers together. You may recall that when referring to data
storage, the term media refers to materials for storing data, such as magnetic
:disks and tape. In network communications, however, media refers to the wires,
cables, and other means by which data travels from its source to its
destination. The most common media for data communication are twisted-pair wire,
coaxial cable, fiber-optic cable, and wireless links.
Twisted-Pair Wire
Twisted-pair wire normally consists of four or eight cop-per strands of
wire, individually insulated in plastic, then twisted around each other in
braided pairs and bound together in another layer of plastic insulation. (There
were originally just two wires, rather than four or eight, hence the name
twisted-pair.) Except for the plastic coating, noth-ing shields this type of
wire from outside interference, so it is also called unshielded twisted-pair (UTP)
wire. Some twisted-pair wire is further encased in a metal sheath and therefore
is called shielded twisted-pair (STP) wire. Figure
7.16 shows what UTP and STP look like.
Indoor wiring for telephones uses twisted-pair wire, so twisted-pair is often
called telephone wire. Because it was readily available and inexpensive,
telephone wire gained early favor as a conduit for data communications. Today,
however, some twisted-pair wire used for communication is made to more demanding
specifications than voice-grade telephone wire.
Sometimes network media are compared by the amount of data they can transmit
each second. The difference between the highest and lowest frequencies of a
transmis-sion channel is known as bandwidth. As more users transmit data over a
network, the bandwidth reduces, thereby slowing down all transmissions.
Bandwidth is expressed in cycles per second (hertz) or in bits per sec-ond.
Twisted-pair wire was once considered a low-band-width media, but networks based
on twisted-pair wires now support transmission speeds up to 150 megabits per
second (Mbps), and even faster speeds are on the horizon.
Coaxial Cable
Coaxial cable, sometimes called coax (pronounced "co-axe"), is widely used for
cable TV and is used in some net-works (however, the connectors are different
for TV and networks). There are two conductors in coaxial cable. One is a single
wire in the center of the cable, and the other is a wire mesh shield that
surrounds the first wire with an insulator in between (see Figure 7.17).
Coaxial cable can carry more data than older types of twisted-pair wiring,
and it is less susceptible to interference from other wiring. However, it is
also more expensive and has become less popular as twisted-pair technology has
improved. Two types of coaxial cable are used with networks: thick and thin.
Thick coax is the older standard and is seldom installed in new networks.
Fiber-Optic Cable
A fiber-optic cable is a thin strand of glass that transmits pulsating
beams of light rather than electric frequencies (see Figure 7.18). When one end
of the strand is exposed to light, the strand carries the light all the way to
the other end--bending around corners with only a minute loss of energy along
the way. Because light travels at a much higher frequency than electrical
signals, fiber-optic cable can easily carry data at more than a billion bits per
second usually 1300 Mbps. Fiber-optic cable is also immune to the
electromag-netic interference that is a problem for copper wire.
The disadvantage of fiber-optic cable is that it is more expensive than
twisted-pair and coax, and it is more difficult to install because it does not
bend around corners as easily. As costs have come down, however, fiber-optic
cable has become increasingly popular, and it is now revolutionizing a number of
communications industries. Telephone and cable television companies, especially,
have been moving from twisted-pair wire and coaxial cables to fiber-optic
cables.
The Network Interface Card, Network Protocols,
and Cabling Specifications
Cables or radio waves may link a network together, but each computer on
the network still needs hardware to control the flow of data. The device that
per-forms this function is the network interface card (NIC). The NIC is a type
of expansion board--a printed circuit board that fits into one of the computer's
expansion slots and provides a port on the back of the PC to which the network
cable attaches. The computer also requires network software, which tells the
computer how to use the NIC.
Both the network software and the NIC have to adhere to a network protocol,
which is a set of standards for communication. A network protocol is like a
language computers use for communicating data. For computers to share data, they
must speak the same language. As is often the case with computer technologies,
protocols are in a continual state of flux. Whenever someone comes up with a new
standard, someone else invents another that does the same job faster and more
reliably. The most common protocols used in networks are TCP/IP, IPX/SPX, and
NetBEUI.
Another specification of a network is to determine the network technology or
cabling equipment used to create a LAN. The most common types of net-work
technology include Ethernet (which also includes Fast Ethernet), Token Ring, and
ARCnet. Each of these is designed for a certain kind of network topology and has
certain standard features.
Ethernet
Currently, Ethernet is the most common network technology used. Ethernet
was originally designed for a bus topology and thick coaxial cable. More recent
implementations have moved to the star topology and twisted-pair wires. Ethernet
requires each computer or workstation on the network to wait its turn to send
data. When a computer needs to send data to another computer or to a peripheral
device, it must first determine whether the network is available. If the network
is in use by another node, the computer waits a tiny fraction of a second and
tries again. If two nodes inadvertently transmit simultaneously, the conflict is
detected and they retransmit one at a time. This approach to network
communication is called CSMA/CD (Carrier Sense Multiple Access/Collision
Detection). As you might guess, when many computers are on an Ethernet network,
access time can become noticeably delayed.
The original implementations of Ethernet, which used coaxial cable, were
called 10Base-5 and 10Base-2. The most popular implementation of Ethernet
currently is called 10Base-T. It uses a star topology and twisted-pair wires and
can achieve transmission speeds up to 10 Mbps.
Fast Ethernet
100Base-T, also known as Fast Ethernet, is available using the same media
and topology as Ethernet, but different network interface cards are used to
achieve speeds of up to 100 Mbps. The 3COM EtherLink XL 10/100 adapter is an
example of a network interface card used to achieve 100 Mbps LAN speeds.
Hewlett-Packard's 100Base-VG competes with 100Base-T. Still other
implementations of Ethernet are pushing transmission speeds even higher.
Token Ring
IBM's network technology is the Token Ring. The controlling hardware in a
Token Ring network transmits an electronic token--a small set of data that
includes an address--to each node on the network many times each second. If the
token is not currently in use, a computer can copy data into the token and set
the address where the data should be sent. The token then continues around the
ring. Each computer along the way looks at the address until the token reaches
the computer with the address that was recorded in the token. The receiving
computer then copies the contents of the token and resets the token status to
empty.
Token Ring networks have the advantage of data traveling through the ring in
one direction in a controlled manner. With this approach, data cannot collide,
so a complex scheme like CSMA/CD is not necessary. However, the network hardware
is not cheap; Token-Ring adapter cards can cost as much as five times more than
other types of network adapters. Token Ring networks once operated at either 4
or 16 Mbits per second, hut as with Ethernet, new tech-nology has pushed the
transmission rate up to 100 Mbits per second.
ARCnet
ARCnet (Attached Resource Computer network) has both a topology and
net-working technology all its own. ARCnet uses either twisted-pair wire or
coax-ial cable, and the star topology is formed with hubs attached to the
network.
NETWORK SOFTWARE
Most of the networking terms you have seen so far--with the exception of
the protocols discussed in the previous section--have referred to hardware. As
with every other part of the computer system, however, there must be software to
control tile hardware. The group of programs that manages the resources on the
net-work is often called the network operating system, or NOS.
The most popular NOS, NetWare, from Novell, can be used to run networks with
different protocols, including Ethernet, Token Ring, and ARCnet. NetWare also
includes support for various hardware platforms, such as Mac, PC, and UNIX nodes
and servers.
One of NetWare's competitors, Banyan's VINES, offers similar flexibility.
Some other NOSs, such as Windows NT Server, DECNet, LANtastic, and AppleShare
are designed to implement specific network protocols on specific types of
machines.
Back to the Taylor Family Free Genealogy Home Page
Updated 21 Nov 2005 Copyright 1999 - 2005 by
John R. Taylor
