Network architecture describes how computer network is arranged and how computer
resources are shared.
There are a number of specialized terms that describes computer network. Some terms often
used with networks are: node, client, server, network operating system, distributed processing
and host computer.
A node is any device that is connected to a network. It could be a computer, printer, or
communication or data storage device.
A client is a node that requests and uses resources available from other nodes. Typically, a

client is a user's microcomputer.
A server is a node that shares resources with other nodes. Depending on the resources
shared, it may be called a file server, printer server, communication server, or database server.
Network operating system likes Windows, it controls and coordinate the activities between
computers on a network. These activities include electronic communication, information, and
resource sharing.
In a distributed processing system, computing power is located and shared at different
locations. [1]This type of system is common in decentralized organizations where divisional
offices have their own computer systems. The computer systems in the divisional offices are
networked to the organization's main or centralized computer.
Host computer is a large centralized computer, usually a minicomputer or a mainframe.
A network may consist only of microcomputers, or it may integrate microcomputers or
other devices with large computers. [2]Networks can be controlled by all nodes working together
equally or by specialized nodes coordinating and supplying all resources. Networks may be
simple or complex, self-contained or dispersed over a large geographical area.
Configuration A network can be arranged or configured in several different ways. The
four principal configurations are star, bus, ring, and hierarchical.
In a star network, a number of small computers or peripheral devices are linked to a central
unit. This central unit may be a host computer or a file server. All communications pass through
this central unit. Control is maintained by polling. That is, each connecting device is asked
whether it has a message to send. Each device is then in turn allowed to send its message. One
particular advantage of the star form of network is that it can be used to provide a time-sharing
system. That is, several users can share resources ("time") on a central computer. The star is a
common arrangement for linking several microcomputers to a mainframe that allows access to an
organization’s database.
In a bus network, each device in the network handles its own communications control. There
is no host computer. All communications travel along a common connecting cable called a bus. As
the information passes along the bus, it's examined by each device to see if the information is
intended for it. The bus network is typically used when only a few microcomputers are to be linked
together. This arrangement is common in systems for electronic mail or for sharing data stored on
different microcomputers. The bus network is not as efficient as the star network for sharing
common resources. (This is because the bus network is not a direct link to the resource.) However,
a bus network is less expensive and is in very common use.
In a ring network, each device is connected to two other devices, forming a ring. There is no
central file server or computer. Message are passed around the ring until they reach the correct
destination. With microcomputers, the ring arrangement is the least frequently used of the four
networks. However, it often is used to link mainframes, especially over wide geographical areas.
These mainframes tend to operate fairly autonomously. They perform most or all of their own
processing and only occasionally share data and programs with other mainframes. A ring

network is useful in a decentralized organization because it makes possible a distributed data
processing system. That is, computers can perform processing tasks at their own dispersed
locations. However, they can also share programs, data and other resources with each other.
The hierarchical network consists of several computers linked to a central host computer,
just like a star network. However, these other computers are also hosts to other, smaller
computers or to peripheral devices. Thus, the host at the top of the hierarchy could be a
mainframe. The computers below the mainframe could be minicomputers, and those below,
microcomputers. The hierarchical network—also called a hybrid network—allows various
computers to share databases, processing power, and different output devices. A hierarchical
network is useful in centralized organizations. For example, different departments within an
organization may have individual microcomputers connected to departmental minicomputers.
The minicomputers in turn may be connected to the corporation’s mainframe, which contains
data and programs accessible to all.
Strategies

Every network has a strategy or way of coordinating the sharing of information
and resources. The most common network strategies are peer-to-peer and client/server systems.
In a peer-to-peer network system nodes can act as both servers and clients. For example, one
microcomputer can obtain files located on another microcomputer and can also provide files to
other microcomputers. A typical configuration for a peer-to-peer system is the bus network.
Commonly used net operating systems are Apple's Macintosh Peer-to-Peer LANs, Novell’s
Netware Lite, and Microsoft's Windows for Workgroups. There are several advantages to using this
type of strategy. The networks are inexpensive and easy to install, and they usually work well for
smaller systems with less than ten nodes. As the number of nodes increases, however, the
performance of the network declines. Another disadvantage is the lack of powerful management
software to effectively monitor a large network's activities. For these reasons, peer-to-peer network
are typically used by small networks.
Client/server network systems use one powerful computer to coordinate and supply services to
all other nodes on the network. This strategy is based on specialization. Server nodes coordinate
and supply specialized services, and client nodes request the services. Commonly used net
operating systems are Novell's Netware, Microsoft's LAN and Windows NT. One advantage of
client/server network systems is their ability to handle very large networks efficiently. Another
advantage is the powerful network management software that monitors and controls the network's
activities. The major disadvantages are the cost of installation and maintenance.

1, term  [tə:m]
n. 学期,术语,名词,期限
v. 称,呼
2, peripheral  [pə'rifərəl]
a. 不重要的,外围的
3, hybrid  ['haibrid]
n. 杂种，混血儿；混合物
adj. 混合的；杂种的
4, dispersed  [di'spə:st]
adj. 散布的；被分散的；被驱散的
v. 分散；传播（disperse的过去分词）