Resource-Aware Energy Efficient Workflow Scheduling in Cloud Infrastructure

Energy-efficient workflow scheduling is a challenging issue of cloud computing paradigm and it is also recognized as an NP-complete problem. During the scheduling, the workflow scheduler allocates the user request to the set of available virtual machines (VMs) which are deployed on cloud servers. However, the schedule length (makespan) and incurred cost for processing the workflow application must remain within the deadline and budget of the user while minimizing the energy usage. This paper proposes a strategy to minimize the overall energy consumption by maximizing the utilization of some VMs while deallocating the VMs that are contributing very less to the workflow execution. The proposed heuristic is simulated thoroughly on a wide range of scientific workflows. The experimental results of the proposed algorithm outperform the existing workflow scheduling heuristic in terms of energy consumption while keeping the cost and makespan within budget and deadline limits.