A predictive approach to improve a fault tolerance confidence level on grid resources scheduling

Abstract

The grid is becoming more attractive and encouraging platform for solving largescale computing intensive problems. In this environment, various geographically distributed resources are logically coupled together and presented as a single integrated resource. Since resources on grid are dynamic and [farar], one of the most important problems in grid environment is design a fault tolerance resources scheduling. Therefore, finding a stable and fault tolerance resource require designing a predictive method that doing this work. Many methods are presented in a few years ago, but in these algorithms, some parameters such as job requirements and clear predictor method are not truly considered and also some methods apply optimistic view in grid scheduling cycle. On the other hand, since many methods use from GIS information to learn about resources, so they cannot powerfully select optimal nodes because the GIS don't cover all information about grid resources. Due to this disadvantage, this paper presents a new approach on fault tolerance mechanisms for the resource scheduling on grid by using Case-Based Reasoning technique in a local fashion. This approach applies a specific structure in order to prepare fault tolerance between executer nodes to retain system in a safe state with minimum data transferring. Certainly, this algorithm increases fault tolerant confidence therefore, performance of grid will be high.