Please note: This previously published IBM Systems Journal paper was reprinted for this special retrospective issue.
Today there is a wide range of choice for configuring the data processing facilities of an organization-ntralized systems, decentralized systems, small computers, and networks of communicating computers-for distributed data processing. This paper considers the factors that relate to organizations and their data processing requirements and to the various possible data processing configurations. Price-performance ratio, organizational needs, and other factors that recommend the flexibility of distributed data processing are discussed in detail. Also discussed are possible distributed data processing architectures, choice criteria, communications, and application and operating system design principles.