计算机科学与探索 ›› 2012, Vol. 6 ›› Issue (6): 495-503.DOI: 10.3778/j.issn.1673-9418.2012.06.002

• 学术研究 • 上一篇    下一篇

使用移动平均线预测云平台服务性能趋势

陈  光+,白晓颖,黄骁飞,李沐洋,周立柱   

  1. 清华大学 计算机科学与技术系,北京 100084
  • 出版日期:2012-06-01 发布日期:2012-06-01

Cloud Performance Trend Prediction by Moving Averages

CHEN Guang+, BAI Xiaoying, HUANG Xiaofei, LI Muyang, ZHOU Lizhu   

  1. Department of Computer Science and Technology, Tsinghua University, Beijing 100084, China
  • Online:2012-06-01 Published:2012-06-01

摘要: 系统性能预测为有预见性的资源调度提供依据,是云平台管理的重要方面。从数据收集、数据处理和预测方法三个方面总结性能预测方法。提出采用移动平均线方法来预测服务性能的长期发展趋势。针对性能小范围波动时会频繁改变预测信号的问题,进一步改进此方法,引入标准差以有效过滤抖动信号。在亚马逊弹性计算云环境下验证了方法的有效性。

关键词: 性能预测, 云计算, 移动平均线

Abstract: Performance prediction is necessary for intelligent resource allocation on the cloud platform. This paper analyzes present predicting methods from three aspects: data collection, data processing and predication methods. Then, it proposes an approach for long-term trend prediction using moving averages method. To better tolerate performance jitter in a small range, it further improves the conventional moving averages method with signal filtering mechanism using standard deviations. An experiment is exercised on Amazon Elastic Compute Cloud platform to illustrate the proposed approach of performance monitoring, analysis and prediction.

Key words: performance prediction, cloud computing, moving averages