Energy-efficient scheduling of real-time cloud services using task consolidation and Dynamic Voltage Scaling

Energy consumption has attracted a lot of attention in the past few years, because energy reduction causes a significant mitigation of the negative impact on the environment along with an operational cost reduction. Energy-efficient task scheduling is an effective technique to decrease the energy consumption in the Cloud Computing Systems (CCSs). In this paper, the problem of scheduling a set of precedence-constrained real-time services onto a set of heterogenous servers is investigated. Each service contains a set of tasks bounded with a specific deadline. The main notion applied in this paper is to employ the consolidation approach along with the Dynamic Voltage Scaling (DVS) technique. The proposed scheduler is developed in three phases. Tasks´ deadlines and a laxity metric are computed for each service according to the corresponding service deadline prior to the main scheduling phase. Afterwards, in order to consolidate the tasks onto the minimum number of servers, the algorithm estimates the required number of servers. Finally, in the last phase, the tasks are scheduled while the DVS technique is applied with considering the tasks´ deadlines. The extensive experimental results clearly demonstrate that the proposed algorithm reduces the energy consumption of a CCS by 14% on average in comparison with beam search algorithm. In addition, it outperforms the non power-aware algorithm by 84%.