Ninth IEEE International Symposium on High-Assurance Systems Engineering (HASE'05)
The Reliable Platform Service: A Property-Based Fault Tolerant Service Architecture
Heidelberg, Germany
October 12-October 14
ISBN: 0-7695-2377-3
The Reliable Platform is a fault tolerant architecture designed to provide a structured but flexible framework for the delivery of dependable services for highly critical applications such as X-by-wire systems. The approach is based on defining a structured hierarchy of critical fault tolerant services with corresponding properties that can be explicitly specified and verified. The architecture also incorporates a comprehensive error model that is inclusive of symmetric and asymmetric (i.e. Byzantine) errors of both a permanent and transient nature. Advanced features include the use of hybrid error recovery algorithms, and node /process level synchronization strategies. The system is capable of managing diverse processes at different levels of severity and with varied failure semantics. The system is dynamically reconfigurable based on error containment regions and on-line diagnosis protocols.
Index Terms:
Fault Tolerant Architecture, Distributed Systems, On-Line Diagnosis, Customizable Fault Effects Model, Middleware
Citation:
Chris Walter, Peter Ellis, Brian LaValley, "The Reliable Platform Service: A Property-Based Fault Tolerant Service Architecture," hase, pp.34-43, Ninth IEEE International Symposium on High-Assurance Systems Engineering (HASE'05), 2005