Optimizing Data-Accessing Energy Consumption for Workflow Applications in Clouds

As more and more non-trivial applications have been deployed in cloud-based systems, the energy consumption of running these applications grow rapidly. Existing studies mainly focus on reducing the CPU-related energy consumption, while ignoring the data-accessing related energy costs. In the paper, we present a novel energy-efficient policy, which is aiming at reducing the data-accessing energy consumption of workflow applications that executed on cloud environments. The proposed policy uses a general energy model to describe the energy consumption of any given workflow. By this application-oriented energy model, a two-phase resource deploying algorithm is implemented, which is capable of generating task scheduling schemes with minimal data-accessing energy consumption. Massive experiments are conducted on a real-world cloud platform, and the results show that the proposed policy outperforms existing approaches in terms of energy efficiency and execution performance.