Details

Autor(en) / Beteiligte

• Mingming Zhang
• Songyun Wang
• Guoquan Yuan
• Yefei Li
• Zhuzhong Qian

Titel

Energy-Efficient Real-Time Task Allocation in a Data Center

Ist Teil von

• 2016 IEEE International Conference on Internet of Things (iThings) and IEEE Green Computing and Communications (GreenCom) and IEEE Cyber, Physical and Social Computing (CPSCom) and IEEE Smart Data (SmartData), 2016, p.680-687

Ort / Verlag

IEEE

Erscheinungsjahr

2016

Quelle

IEEE/IET Electronic Library (IEL)

Beschreibungen/Notizen

• Energy saving is one of the most important issues in the design of data centers. Dynamic Voltage and Frequency Scaling (DVFS) is recognized as an effective technique for achieving the tradeoff between system performance and energy saving. Many previous works addressed the DVFS-based energy-aware task allocation using total workload (execution time) of tasks, but it is extremely difficult to obtain in some scenarios. In this paper, we investigate the task allocation to minimize the energy consumption of clusters based on a new common task model, which present task requirements using the minimum frequency instead of execution time. We propose Energy Consumption Ratio (ECR) to evaluate the energy consumption of executing a task under different frequencies, which are constants to processors. Thus, the energy efficient task allocation problem can be converted into minimizing the total ECR. We proved this optimization problem is analogous but more difficult than Variable Size Bin Packing. We propose two lower-order polynomial-time heuristics, allocating tasks among servers. Plenty of experiments show that our algorithms outperform some other energy-aware strategies on energy saving. And we also proved our task allocation mechanism do achieve significant energy saving in a real testbed.