Eco-friendly Dynamic Task Scheduling for Regional Data Center

The resource requirement of service request of users are commonly uncertain. Thus, it is becoming increasingly inefficient to aggregate all data required for computation at a central data center. Recently, to achieve the better reliability and performance, the system has been largely depending on the regional data centers. Instead, a more recent trend is to distribute computation to meet data locality, thus reducing the resource requirement (e.g., CPU, RAM, bandwidth) while improving performance. Consequently, new challenges are emerging in task scheduling, where each task runs across regional data centers and, requiring coordination among regional data centers located in different edge, and stronger monitoring mechanism. Hereby, this paper proposes an eco-friendly dynamic task scheduling method (EcoRS) based on power prediction and threshold based queuing model for regional data centers to improve the quality of services (QoS), not merely should the traditional standards such as energy and cost be satisfied, but particular emphasis should be laid on some extended standards such as environmental footprint, while maintaining service level agreement (SLA).