In the 1980s a new architecture arose: the client-server architecture. Two technological advances in information technology made the purchase of computers affordable for almost every organization.
Firstly, the so-called mini-computers or servers were invented. These computers were not as powerful as the traditional mainframes, but they were able to serve several hundreds of users. Competitors of IBM joined forces and developed the new operating system Unix especially for these servers.
Secondly, IBM introduced the personal computer (or: PC) on the market. IBM spared neither money nor effort in their famous Charlie Chaplin campaign, and the PC quickly penetrated the office market and replaced the dumb terminals that were typical for the mainframe architecture.
Client-server architecture is mainly based on the expectation that all elements of a physical architecture are becoming cheaper and more powerful, which enables the distribution of elements over centralized and decentralized locations. Client-server architecture is therefore also known as a distributed architecture. Because of the distribution of elements, performance bottlenecks can be resolved and network traffic, which was still relatively expensive in the last century, can be limited.
In client-server architecture the strict separation between centralized and decentralized elements no longer exists. In this architecture, the client is the user’s workstation. In general, this is a powerful PC with a large amount of internal memory, own disk space, a fast CPU, a graphical monitor, a printer and a keyboard. A PC with these powerful specifications is often called a fat client. The server is a powerful computer that offers large-scale data storage to the clients. The server also has internal memory and a CPU, but these are used only for the fast storage and retrieval of data, and not for computations for the users. he server can be located centrally, but architectures with a server on each location also exist.
In these cases, the servers regularly exchange information with the central server. Between clients and servers, a network is installed. This can either be a cheap local area network (or: LAN) within a location, or a more expensive wide area network (or: WAN) between locations.
When an ERP system is installed on client-server architecture, the ERP elements are distributed over the physical IT architecture. In Figure 4.4 an ERP system is mapped on client-server architecture. The ERP database is installed on the disk space of a server. It is possible to store local data on a local server, while organization-wide data are stored on a central server. In this case, a distributed ERP database is used, and a replication mechanism is required to synchronize the common data.