Multiqueue Scheduling of Heterogeneous Tasks With Bounded Response Time in Hybrid Green IaaS Clouds

Cost-effective task scheduling is an important operation in green infrastructure-as-a-service clouds (GICs) as the energy consumed by users' tasks is drastic. The irregular task arrival forces private GIC to adopt hybrid clouds to outsource some tasks to dynamic and reliable virtual machines (VMs) of public external clouds. However, temporal differences in revenue, electricity prices, wind and solar energy, and VM running prices of public external clouds make it difficult to dispatch all tasks in a cost-effective way while satisfying users' specified response time constraints. Unlike existing methods, we propose a multiqueue scheduling (MQS) method that investigates such temporal differences in hybrid GICs (HGICs). Specially, this work first gives mathematical relations between rejected tasks and service rates of servers in private GIC. In each iteration of MQS, this paper formulates a profit maximization problem for HGIC and solves it by a novel meta-heuristic optimization method by combing simulated annealing, particle swarm optimization, and genetic algorithm. Trace-driven experiments based on real-life data demonstrate that profit and throughput of MQS are larger than typical task scheduling algorithms while meeting tasks' response time constraints.