Multi-Objective Task Scheduling Optimization in Cloud Computing: An Appraisal

Numerous task scheduling problems encountered in cloud computing (CC) environment cannot actually be formulated as a mono-objective problem. Hence, metaheuristic multi-objective (MO) scheduling algorithms are now being proposed to be a potential solution. This appraisal investigated the existing MO task scheduling algorithms that optimize task scheduling in Infrastructure as a Service (IaaS) cloud. However, task scheduling algorithms are drawn from journals and conference proceedings for analysis. Performance metrics for cloud task scheduling were used to evaluate the quality of performance attributes addressed by these algorithms. The percentage of MO Quality of Service (QoS) attributes (Execution cost/Energy consumption 5%, Response time/throughput 5%, Execution time/cost 40%, Execution time/Load balancing 5%, Execution time/Energy consumption 10%, Execution cost/Makespan 25%, Execution time/Response time 5%, Execution cost, Makespan, Energy consumption/Fault tolerance 5%) addressed by existing algorithms is reported, while comparison based on experimental tools (ClodSim 42%, Matlab 33%, GridSim 10%, Wien2k and AstroGrid 5%, Java 5%, MetteroAG 5%) as used by the existing researchers were also reported. Percentage results showed, most existing MO scheduling algorithms minimized QoS attributes like task execution time, cost, and makespan without emphases on the scalability, reliability, energy consumption and fault tolerance. With the adopted method for evaluating the existing algorithms, findings of this research will help researchers with further research directions on how to analyze different techniques pointing at the main research problem.