作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程

• 先进计算与数据处理 • 上一篇    下一篇

大数据处理中基于热感知的能源冷却技术

金伟林,陈国顺   

  1. (浙江财经大学东方学院,杭州310018)
  • 收稿日期:2014-05-19 出版日期:2015-04-15 发布日期:2015-04-15
  • 作者简介:金伟林(1975 - ),男,工程师、硕士,主研方向:大数据处理;陈国顺,高级工程师、硕士。

Energy Cooling Technology Based on Thermal-aware in Big Data Processing

JIN Weilin,CHEN Guoshun   

  1. (East College,Zhejiang University of Finance & Economics,Hangzhou 310018,China)
  • Received:2014-05-19 Online:2015-04-15 Published:2015-04-15

摘要: 大数据极速发展使超大型大数据分析平台不断涌现,导致能源成本急剧上升。为保证服务器的热可靠性,提出一种以数据处理为中心的能源冷却成本技术。该技术考虑了服务器不均衡热力特性、热力稳定性负载阈值差异以及集群大数据语义差异等,对文件进行主动式热感知布局,从而在不影响性能的前提下降低冷却能源成本,保证大数据分析集群的热可靠性。基于Yahoo 公司一个月的真实大数据分析对该技术进行评估,实验结果表明,该技术可使冷却成本下降42% ,总体性能是当前无关冷却技术的9 倍。

关键词: 大数据, 热感知, 热可靠性, 服务器, 能源冷却成本, 集群

Abstract: Explosion in big data has led to a surge in extremely large scale big data analytics platforms,resulting in burgeoning energy costs. In order to ensure the thermal-reliability of the servers,this paper proposes a data-centric technology for reducing cooling energy costs. It considers the uneven thermal-profile of the servers,the differences in their thermal-reliability-driven load thresholds,and differences in the data-semantics of the big data placed in the cluster. Based on this knowledge,the proposed technology does proactive,thermal-aware file placement,which ensures to reduce cooling energy costs and ensures thermal-reliability in the big data analytics cluster without any performance impact. Evaluation results with one-month long real-world big data analytics production traces from Yahoo Experimental results show that the technology reaches 42% reduction in the cooling energy costs and 9x better performance than the state-of-the-art dataagnostic cooling techniques.

Key words: big data, thermal-aware, thermal-reliability, server, energy cooling costs, clusters

中图分类号: