A fault-tolerant architecture for grid system

Abstract

Support for the development of fault-tolerant applications has been identified as one of the major technical challenges to address for the successful deployment of computational grids. Reflective systems provide a very attarctive solution for the problem. But in the traditional systems, all the binding relationships between the function part and the non-function part of the application are fixed. A new reflective architecture which uses events to achieve dynamic binding is proposed in the article. The two best known fault-tolerance algorithms are mapped into the architecture to set forth the dynamic binding protocol. © Springer-Verlag 2004.