Supercomputer
Author:
Keywords:
Science & Technology, Technology, Computer Science, Hardware & Architecture, Computer Science, Theory & Methods, Computer Science
Abstract:
Industrial embedded HPC applications from different application domains have many common requirements with respect to fault tolerance. The EFTOS framework provides a flexible and parametrisable set of tools from which the application developer can select to make the embedded application more fault-tolerant This framework consists of a fault tolerance backbone to which a set of tools can be hooked, These tools provides generic, tailored or application-specific detection and recovery functions. The backbone co-ordinates detection, diagnosis and recovery actions to bring the application back into a consistent state. The integration of this framework in the image processing module of a postal automation system, and in the sequence controller of a High Voltage Substation of an energy distribution network, ensures the industrial usefulness and the flexibility of the approach, while the guideline of portability allows efficient reuse of the framework.