earticle

영어

Recent advancements in cloud computing have significantly increased its importance across various sectors. As sensors, devices, and customer demands have become more diverse, workloads have become increasingly variable and difficult to predict. Cloud providers, connected to multiple physical servers to support a range of applications, often overprovision resources to handle peak workloads. This approach results in inconsistent services, imbalanced energy usage, waste, and potential violations of service level agreements. In this paper, we propose a novel engine equipped with a scheduler based on the Hot-Threshold concept, aimed at optimizing resource usage and improving energy efficiency in cloud environments. We developed this engine to employ both proactive and reactive methods. The proactive method leverages workload estimate-based provisioning, while the reactive Hot-Cold Scheduler consists of a Predictor, Solver, and Processor, which together suggest an intelligent migration flow. We demonstrate that our approach effectively addresses existing challenges in terms of cost and energy consumption. By intelligently managing resources based on past user statistics, we provide significant improvements in both energy efficiency and service consistency.