Definition: centralized processing

Processing performed in one computer or in a cluster of coupled computers in a single location. Access to the computer is via "dumb terminals," which send only input and receive output or "smart terminals," which add screen formatting. All data processing is performed in the central computer.

Centralized processing evolved from the first computers in the 1950s, which were stand-alone with all input and output devices in the same room. Starting in the 1960s, terminals were added throughout a company's headquarters and branch offices, and each terminal was wired into the central machine, whether local or via a remote communications method.

The Terminals Became PCs
As personal computers and local area networks (LANs) proliferated in the 1980s and 1990s, terminals were replaced with PCs, which were made to function like a terminal with the addition of hardware and software. Users could run their own applications and still access the central computer as required (see terminal emulation).

By the 1990s, "client/server" architecture began to replace terminal access. Client PCs communicated with network servers via the local network rather than terminals hardwired to a central machine. Although there are still hardwired terminals in use, the predominant technology today is the local area network (LAN). See client/server.

Thin Clients
The terminal-based centralized architecture is hardly defunct. Harking back to the early days, options within Windows and Unix/Linux servers, as well as software from companies such as VMware and Citrix, enable applications to be run in network servers with the user's computer turned once again into a terminal. Now known as "thin client" computing rather than centralized processing, it is, however, the same concept (see thin client).

Isn't Cloud Computing Centralized?
Yes, it certainly can be. Cloud computing is a type of centralized computing, but it is not typically identified as such. Software as a service (SaaS), the original cloud computing paradigm, implies browser-based interaction with a website and processing that takes place in servers in a datacenter. Whether users are interacting with one server or multiple servers or where the servers happen to be located is generally unknown to cloud computing customers. See cloud computing.