Energy Efficient Resource Scheduling in Cloud Computing Based on Task Arrival Model

High energy consumption has become a vital bottleneck restricting the development of cloud computing. Most of the current resource management frameworks focus on the scheduling module but fail to consider the burstiness of workloads adequately. In this paper, we present a resource management framework based on the arrival model switching mechanism to optimize the energy efficiency of cloud data centers. We analyze cloud task characteristics and propose two types of task arrival models. The arrival model based on the Poisson process is for common scenarios, and the grey model in traffic burst (GMTB) is for bursty scenarios. An anomaly detection module is introduced to detect the abnormal events and determine whether the model needs to be switched. Finally, we propose an integrated virtual machine scheduling policy to balance the energy consumption and service level agreement.