Method for Choosing a Balanced Set of Fault Tolerance Techniques for Distributed Computer Cystems

Abstract

In the paper we consider a method for a reliability allocation problem (RAP) of distributed computer systems (DCS) under cost constraints. In this problem we maximize reliability of DCS under constraints of system cost. The article describes considered fault tolerance mechanisms. The mathematical formulation of RAP is provided. RAP is widely discussed in the literature. A detailed description of the method is ensured. The applied method is an evolutionary algorithm with an adaptive logic control procedure. The adaptive logic control procedure analyzes the results of evolutionary algorithm work in each generation and, based on this information, adjusts parameters. The key feature of the proposed method is the use of an adaptive hybrid genetic algorithm. The results of experiments with the implemented method are presented. This method was implemented as a pilot system which works in cooperation with DYANA simulation environment. Finally, future plans for the development of the presented method and tools are briefly described.

Keywords

• dependability
• fault tolerance techniques
• genetic algorithm
• computer systems
• reliability
• reliability allocation problem
• reliability-redundancy allocation problem
• evolutionary algorithm