A decentralized system-known in the database world as a distributed system can be highly responsive to differences in data gathering, storage, and access. It can adjust to differences in user psychology-between, say, a multinational corporation's employees in individualistic Greece and in disciplined Japan. It can also adapt to the management styles of strong managers in different locations．

An organization that prefers homogeneity and top-down control will naturally choose a centralized database system. It may also prefer the hierarchical data model．

By contrast, an organization that prefers local improvisation and free wheeling may well choose distributed database system. It may also choose the network data model，which is well suited to searching and updating data in a distributed system.

Communication between distributed commutinies of computer is required for many reasons．At a national level, for example, computers located in different parts of the country use public communication services to exchange electronic messages（mail） and to transfer files of information from one computer to another: Similarly, at a local level within, say, a single building or establishment, distributed communities of computer-based workstations use local communication networks to access expensive shared resources-for example, printers, copiers, disks and tapes, etc．-that are also managed by computers．Clearly, as the range of computer-based products and associated public and local com-medication networks procreate, computer-to-computer communication will expand rapidly and ultimately dominate the field of distributed system.

Historically, computers were so expensive that most large organizations did all their data processing on a single, centralized machine. While very efficient for such tasks as payroll and generating accounting reports, the centralized approach was very useful to those who needed a quick response to a unique, local problem.

With today's inexpensive micros and minis, there is no reason why a branch office, the engineer-in department, or any other group needing computer support cannot have its own computer. By linking these remote machines to a centralized computer via communication lines, local activity can be monitored and coordinated. This approach is called distributed data processing.

1, arbitrary  
a. 任意的,恣意的,专制的

2, homogeneity  [,hɔməudʒe'ni:əti, ,həu-]
n. 同质；同种；同次性（等于homogeneousness）

3, improvisation  [,imprəvai'zeiʃən]
n. 即兴创作；即席演奏

4, procreate  ['prəukrieit]
vi. 生育（子女）；产生
vt. 生殖

5, payroll  ['peirəul]
n. 工资单