EFFICIENT QOS BASED SCHEDULING OF FINE-GRAINED TASKS ON GLOBAL GRIDS

Abstract

Grid Computing is a new, and emerging technology in the fields of scientific, and engineering, and as well as in commercial, and industrial enterprises. This growing technology facilitates the conduct of virtual organization by bringing together appropriate, and effective human, information, and computing resources for tackling highly complex, and multidisciplinary projects. Although Grids have been extensively used for executing applications with large compute-intensive jobs, there exist several applications with a large number of lightweight jobs. The overall processing undertaking of these applications involves high overhead time and cost in terms of job transmission to and from Grid resources and job processing at the Grid resources. Job grouping-based scheduling systems dynamically assemble the individual fine-grained jobs of an application into a group of jobs, and send these coarse-grained jobs to the Grid resources, thus by reducing the overheads associated. But while grouping and scheduling, there is need for taking QoS factors such time limits or cost limits and optimizations factors of each job and target resource characteristics in to consideration for overall effective performance of the Grid applications. We solve this problem by presenting a grouping and scheduling strategy that performs dynamic job grouping activity at runtime taking QoS factors specified in the each job and the target resource characteristics into consideration for processing the jobs in the specified deadlines with the specified optimization . We propose three grouping and scheduling algorithms (i) simple cost optimized algorithm (ii) Time Constrained time optimized algorithm and (iii)Budget constrained time optimized algorithm We also evaluate these algorithms with varying the different input parameters such as job size , number of jobs, arrival rate of those jobs and verify the output parameters such as total processing time , total processing cost , percentage of jobs with in the deadline. All the algorithms are simulated in Java with GridSim simualtor library.