Effective fault identification for Optimized fault mitigation in cloud

Abstract

Cloud computing is well known as a technology for providing scalable newlinepay-per-use on-demand service on a high availability basis. High performance newlinesystem requires robust infrastructure with load balancing mechanism for newlinesecuring high reliability. The system needs to have the execution process newlinecompleted without any disruption. To cater to this objective, a fault tolerance newlinemodule is a critical feature in the system. newlineThe fault tolerance represents the ability of a system to continue newlineexecuting processes and providing services without interruption in spite of any newlinefailure or the fault occurrence of any components. Mission critical applications newlineand life-support systems demand high fault tolerance in order to provide newlinesustained services as required. Power failure, disk failure, process failure and newlineothers are overcome in the fault tolerant system by back up resources, disk newlinemirroring and redundant processes. newlineEnhanced Fault Identification and Optimal Task Prediction (EFIOTP) newlinealgorithm covers adaptive task scheduling and on-demand pay-per-use service newlinerenderings in a cloud environment with the minimal execution time of the tasks newlineusing efficient task scheduling. The process complies with the performance newlineparameters on defined time constraints. In the proposed approach, scientific newlineworkflow applications run the scheduled tasks with the defined fault tolerance newlineeffectively. VM migration process is used as a part of heuristic technique for newlinereduction in time and the cost overheads averting VM triggered delays. Fault newlineAware Dynamic Resource Manager (FADRM) handles application level fault newlinetolerance in automatic and transparent application recovery. newline