| Summary: | Grid technologies are established to share the large-scale heterogeneous resources over multiple administrative domains for processing the application. In these technologies, the grid scheduling problem is crucial that must be solved in order to achieve multiple objectives within different stakeholders (end-users, owner resources and administrators) preferences. The aim of this research is to design and implement the Adaptive Intelligent Grid Scheduling System (AIGSS) in order to achieve multiple objectives named Makespan Time, Grid Efficiency and Total delayed jobs. The popular meta-heuristic algorithms, namely Ant Colony Optimization (ACO) and Tabu Search (TS) algorithms are proposed and developed to maintain the selecting appropriate grid resource to execute each job within the different job inter-arrival times and grid resources. Additionally, the clustering technique named Fuzzy C-Means (FCM) algorithm is proposed for clustering the groups of grid resources as well as jobs based on the degree of characteristic similarity. Moreover, a popular discrete event simulation tool, namely, GridSim toolkit and Alea simulation, is extended by developing the service modules on top of it. Therefore, the experiment is simulated as realistic grid environment in order to measure the proposed system. The experimental results show that the AIGSS provides reasonable multiple objectives to stakeholders within different job interarrival times and machines in grid system. In addition to the experimental results, the proposed system performs better than the other algorithms for different goals of each stakeholder. The performance of AIGSS is compared with the common and heuristic algorithms such as First-Come-First-Serve (FCFS) with Optimization, Earliest Deadline First (EDF), Minimum Tardiness Earliest Due Date (MTEDD), Minimum Completion Time (MCT), Opportunistic Load Balancing (OLB), MIN-MIN, Hill Climbing, EASY Backfilling, Simulated Annealing (SA), and Tabu Searching (TS).
|
|---|