Optimized task scheduling based on hybrid symbiotic organisms search algorithms for cloud computing environment

Abstract

In Cloud Computing model, users are charged according to the usage of resources and desired Quality of Service (QoS). Task scheduling algorithms are responsible for specifying adequate set of resources to execute user applications in the form of tasks, and schedule decisions of task scheduling algorithms are based on QoS requirements defined by the user. Task scheduling problem is an NP-Complete problem, due to the NP-Complete nature of task scheduling problems and huge search space presented by large scale problem instances, many of the existing solution algorithms incur high computational complexity and cannot effectively obtain global optimum solutions. Recently, Symbiotic Organisms Search (SOS) has been applied to various optimization problems and results obtained were found to be competitive with state-of-the-art metaheuristic algorithms. However, similar to the case other metaheuristic optimization algorithms, the efficiency of SOS algorithm deteriorates as the size of the search space increases. Moreover, SOS suffers from local optima entrapment and its static control parameters cannot maintain a balance between local and global search. In this study, Cooperative Coevolutionary Constrained Multiobjective Symbiotic Organisms Search (CC-CMSOS), Cooperative Coevolutionary Constrained Multi-objective Memetic Symbiotic Organisms Search (CC-CMMSOS), and Cooperative Coevolutionary Constrained Multi-objective Adaptive Benefit Factor Symbiotic Organisms Search (CC-CMABFSOS) algorithms are proposed to solve constrained multi-objective large scale task scheduling optimization problem on IaaS cloud computing environment. To address the issue of scalability, the concept of Cooperative Coevolutionary for enhancing SOS named CC-CMSOS make SOS more efficient for solving large scale task scheduling problems. CC-CMMSOS algorithm further improves the performance of SOS algorithm by hybridizing with Simulated Annealing (SA) to avoid entrapment in local optima for global convergence. Finally, CC-CMABFSOS algorithm adaptively turn SOS control parameters to balance the local and global search procedure for faster convergence speed. The performance of the proposed CC-CMSOS, CC-CMMSOS, and CC-CMABFSOS algorithms are evaluated on CloudSim simulator, using both standard workload traces and synthesized workloads for larger problem instances of up to 5000. Moreover, CC-CMSOS, CC-CMMSOS, and CC-CMABFSOS algorithms are compared with multi-objective optimization algorithms, namely, EMS-C, ECMSMOO, and BOGA. The CC-CMSOS, CC-CMMSOS, and CC-CMABFSOS algorithms obtained significant improved optimal trade-offs between execution time (makespan) and financial cost (cost) while meeting deadline constraints with no computational overhead. The performance improvements obtained by the proposed algorithms in terms of hypervolume ranges from 8.72% to 37.95% across the workloads. Therefore, the proposed algorithms have potentials to improve the performance of QoS delivery.