An Agent Oriented Proactive Fault-Tolerant Framework for Grid Computing

Abstract

Because of computational grid heterogeneity, scale and complexity, faults become likely. Therefore, grid infrastructure must have mechanisms to deal with faults while also providing efficient and reliable services to its end users. Existing fault-tolerant approaches are inefficient because they are reactive and incomplete. They are reactive because they only deal with faults when they take place; they are incomplete because they only deal with certain types of faults. Proactive approaches increase efficiency by reducing the cost and time of operations and network resource usage by maintaining the state of executing applications and resuming operation when rescheduled. This paper presents an agent oriented, fault-tolerant grid framework where agents deal with individual faults proactively. Agents maintain information about hardware conditions, executing process memory consumption, available resources, network conditions and component mean time to failure. Based on this information and critical states, agent can improve the reliability and efficiency of grid services.