2014 IEEE 6th International Conference on Cloud Computing Technology and Science (CloudCom) (2014)
Dec. 15, 2014 to Dec. 18, 2014
The broad adoption of cloud services led to an increasing concentration of servers in a few data centers. Reports estimate the energy consumptions of these data centers to be between 1.1% and 1.5% of the worldwide electricity consumption. This extensive energy consumption precludes massive CO2 emissions, as a significant number of data centers are backed by "brown" power plants. While most researchers have focused on reducing energy consumption of cloud data centers via server consolidation, we propose an approach for reducing the power required to execute urgent, CPU-intensive Bag-of-Tasks applications on cloud infrastructures. It exploits intelligent scheduling combined with the Dynamic Voltage and Frequency Scaling (DVFS) capability of modern CPU processors to keep the CPU operating at the minimum voltage level (and consequently minimum frequency and power consumption) that enables the application to complete before a user-defined deadline. Experiments demonstrate that our approach reduces energy consumption with the extra feature of not requiring virtual machines to have knowledge about its underlying physical infrastructure, which is an assumption of previous works.
Energy consumption, Virtual machining, Servers, Heuristic algorithms, Scheduling, Processor scheduling, Computational modeling
R. N. Calheiros and R. Buyya, "Energy-Efficient Scheduling of Urgent Bag-of-Tasks Applications in Clouds through DVFS," 2014 IEEE 6th International Conference on Cloud Computing Technology and Science (CloudCom)(CLOUDCOM), Singapore, Singapore, 2014, pp. 342-349.