Scalable Fault Tolerant System Design Using Heterogeneous Components

An integrated system design approach will be pursued which considers the important aspects for scalable fault tolerant computing. An architecture approach, using selective separation of fault tolerant services from application and COTS components will provide transparency and interoperability among heterogeneous components. Real-time services and policy driven configuration and control will be supported. The key service goals are high performance, scability, fault tolerance and flexibility of embedding. The key implementation goals are use of standard components and interfaces, and support common programming languages and ADa 95. Three forms of multiprocessing will be supported in additions to the use of the architecture in an embedded environment. Concepts and algorithms which can be formally verified will be developed and used. System test capabilities will be addressed to provide reliable information on the collective health of the system elements. Most importantly, the method will enable the construction of system architecture that are scalable in terms of performance and fault tolerance, and suitable for applications ranging from highly available to ultradependable mission critical computing.