A model to support consistency and availability in distributed system architectures

An alternative to inherently centralized control models and time-variant data models for distributed computing systems is proposed. This model is characterized by a decentralized control model and a data model based on ideas from functional programming. The control model aids availability and extensibility of the system management. The data model aids availability of particular objects. It is shown that the data model significantly reduces the severity of the consistency problems normally encountered in distributed systems.<>