Energy and cost aware scheduling with batch processing for instance-intensive IoT workflows in clouds

Abstract Cloud computing is a suitable platform to execute various applications. At the same time, it should not only provide QoS such as high throughput but also achieve relevant criteria such as efficient power consumption and appropriate execution cost. To address this challenge, we present an energy and cost aware algorithm for scheduling instance-intensive IoT workflows with batch processing in clouds, which is named ECIB and aimed to improve energy efficiency and reduce execution cost while meeting the deadline requirements. Specifically, we propose a prediction based strategy to guide the management of resources by the historical data and CPU usage prediction results of physical machines. Then, we propose two strategies to scale up or scale down the virtual machine resources to optimize the energy consumption for the cloud data centers. In addition, we adopt a batch processing strategy to merge some activity instances of the same type so as to reduce execution cost for the cloud users and improve resource utilization for the cloud data centers. The effectiveness of the ECIB algorithm is evaluated by extensive experiments using CloudSim and four types of instance-intensive IoT workflow applications.