Design and Development of a fuzzy logic based grid scheduler system

Abstract

Grid infrastructure is a highly distributed environment. Availability, usage and cost policies vary depending on the particular user, time, priorities and goals. The management of resources and application scheduling in such a large-scale distributed environment is a complex task. These highly scalable systems are designed to meet the large computational demands of many users from scientific and business orientations. Scheduling is an important design issue in Grid computing. Grid scheduling is defined as the process of making scheduling decisions involving allocating jobs to resources over multiple administrative domains. Grid computing systems aims to enable sharing, selection and aggregation of wide variety of systems including supercomputers, storage systems and other distributed systems in different geographical locations. Current scheduling algorithms are based on single processor systems. First Come First Serve (FCFS), Earliest Deadline First (EDF) and Easy Backfilling (EASY) are common algorithms implemented in Grid middleware's. This dissertation aims to address the problems related to Grid scheduling. This is achieved by designing and developing a fuzzy logic based grid scheduling system to enable the optimum use of the Grid system. The primary contribution of this research work was allocation of right job to the most suitable system by applying the knowledge based capability of fuzzy inference engine of MATLAB. This research compared the proposed model with various well use Grid scheduling algorithms and showed a good performance against these legacy algorithms. This was achieved through the use of a simulation environment defined in ALEA simulation system. Hence doing so, the delay processing jobs are eliminated and a much optimal response time was shown for the jobs submitted to the Grid system. On average the response time was reduced by 15% and 10% against FCFS and EDF respectively.