面向按需供给的资源需求滤波估算方法

黄翔; 陈伟; 宋云奎; 陈志刚

doi:10.3724/SP.J.1004.2014.00942

面向按需供给的资源需求滤波估算方法

doi: 10.3724/SP.J.1004.2014.00942

1.
中国能源建设集团广东省电力设计研究院广州 510663;
2.
中山大学信息科学与技术学院广州 510006;
3.
中国科学院软件研究所北京 100190

基金项目:

国家自然科学基金（61272013），广东省自然科学基金（S2013040011941）资助

详细信息

作者简介:
陈伟中国科学院软件研究所助理研究员. 2013 年获得中国科学院博士学位.主要研究方向为网络分布式计算，软件工程.E-mail：wchen@otcaix.iscas.ac.cn

计量
- 文章访问数: 1745
- HTML全文浏览量: 54
- PDF下载量: 677
- 被引次数: 0
出版历程
- 收稿日期: 2013-05-22
- 修回日期: 2013-08-22
- 刊出日期: 2014-05-20

Filter Based Resource Demand Estimation for On-demand Provision

1.
Guangdong Electric Power Design Institute, China Energy Engineering Group, Guangzhou 510663;
2.
School of Information Science and Technology, Sun Yat-Sen University, Guangzhou 510006;
3.
Institute of Software, Chinese Academy of Sciences, Beijing 100190

Funds:

Supported by National Natural Science Foundation of China (61272013) and Guangdong Natural Science Foundation of China (S2013040011941)

摘要

摘要: 随着按需供给资源使用模式的推广，软件的资源需求已成为资源优化控制的重要属性.监测和估算是目前常用的资源消耗获取方法，但监测工具难以在运行时准确度量短任务的资源需求，回归分析方法又因受到多元共线性和不确定性因素的影响，导致其取值精度下降.本文提出了一种基于Kalman滤波的资源需求估算方法.该方法建立了可度量属性集与不可度量的资源需求间的关联，并利用滤波过滤度量过程中的噪声，达到降低估算误差的目的.基准测试的结果表明，通过合理的设置滤波参数，本方法能够快速逼近真实值，且平均误差小于8%.
- 按需供给 /
- 资源需求 /
- 滤波 /
- 估算
Abstract: As the development of demand resource provision, resource demands of software is becoming one of the most important attributes of resource management. Measurement and estimation are widely used in fetching the demands. However, it is hard to measure the short job's resource demands by current measurement tools, and the regression methods suffer from the well-studied problem of multicollinearity. Therefore, the estimated results are not confident. In order to improve the estimation precision, we propose a Kalman filter based approach, which can predict the unobservable attribute by observable attributes, and filter the noise existing in the measurement. At last, we test our approach with a benchmark and compare the relative errors, which can demonstrate that with the reasonable parameters, our approach can get close to the real demands quickly, and get the estimated value with the mean error less than 8%.
- On demand /
- resource demand /
- flter /
- estimation

HTML全文

参考文献(26)

[1]	Jordan M, Czajkowski G, Kouklinski K, Skinner G. Extending a J2EE server with dynamic and flexible resource management. In: Proceedings of the 2004 International Conference on Middleware. Toronto, Canada: ACM, 2004. 439-458
[2]	Binder W, Hulaas J. A portable CPU-management framework for Java. IEEE Internet Computing, 2004, 8(5): 74-83
[3]	Hulaas J, Kalas D. Monitoring of resource consumption in Java-based application servers. In: Proceedings of OpenView University Association 10th Workshop. Geneva, Switzerland: University of Geneva, 2003. 1-6
[4]	Islam S, Keung J, Lee K, Liu A. Empirical prediction models for adaptive resource provisioning in the cloud. Future Generation Computer Systems, 2012, 28(1): 155-162
[5]	Kalbasi A, Krishnamurthy D, Rolia J, Dawson S. DEC: service demand estimation with confidence. IEEE Transactions on Software Engineering, 2012, 38(3): 561-578
[6]	Teunissen P J G, Khodabandeh A. BLUE, BLUP and the Kalman filter: some new results. Journal of Geodesy, 2013, 87(5): 461-473
[7]	You Ke-You, Xie Li-Hua. Survey of recent progress in networked control systems. Acta Automatica Sinica, 2013, 39(2): 101-118(游科友, 谢立华. 网络控制系统的最新研究综述. 自动化学报, 2013, 39(2): 101-118)
[8]	Kalman R E. A new approach to linear filtering and prediction problems. Journal of Basic Engineering, 1960, 82(1): 35-45
[9]	Tao Gui-Li, Deng Zi-Li. Self-tuning fusion Kalman filter with unknown parameters and its convergence. Acta Automatica Sinica, 2012, 38(1): 109-119(陶贵丽, 邓自立. 含未知参数的自校正融合Kalman滤波器及其收敛性. 自动化学报, 2012, 38(1): 109-119)
[10]	Lazowska E D, Zahorjan J, Graham G S, Sevcik K C. Quantitative System Performance: Computer System Analysis Using Queueing Network Models. Upper Saddle River, NJ, USA: Prentice-Hall, 1984
[11]	Hou Zhen-Wei, Fang Hai-Tao. L 2-stability of discrete-time Kalman filter with random coefficients under incorrect covariance. Acta Automatica Sinica, 2013, 39(1): 43-52(周振威, 方海涛. 在不准确方差下带随机系数矩阵的卡尔曼滤波稳定性. 自动化学报, 2013, 39(1): 43-52)
[12]	Rao J, Wei Y D, Gong J Y, Xu C Z. QoS guarantees and service differentiation for dynamic cloud applications. IEEE Transactions on Network and Service Management, 2013, 10(1): 43-55
[13]	Malek O, Venetsanopoulos A, Alamgir L, Alirezaie J, Krishnan S. A discrete-time convergence model for proliferation-able stem cell and its estimation using Kalman filter. Journal of Bioengineer and Biomedical Sciences, 2013, 3(1): 1-10
[14]	You K Y, Xie L H. Kalman filtering with scheduled measurements. IEEE Transactions on Signal Processing, 2013, 61(6): 1520-1530
[15]	Bard Y, Shatzoff M. Statistical methods in computer performance analysis. Current Trends in Programming Methodology. New Jersey: Prentice-Hall, 1978, 3: 1-51
[16]	Han R, Guo L, Ghanem M M, Guo Y K. Lightweight resource scaling for cloud applications. In: Proceedings of the 2012 Cluster, Cloud and Grid Computing. Ottawa, CA: IEEE, 2012, 644-651
[17]	Pacifici G, Segmuller W, Spreitzer M, Tantawi A. CPU demand for web serving: measurement analysis and dynamic estimation. Performance Evaluation, 2008, 65(6-7): 531-553
[18]	Simon Spinner. Evaluating Approaches to Resource Demand Estimation [Master dissertation], University of the State of Baden-Wuerttemberg and National Laboratory of the Helmholtz Association, Germany, 2011
[19]	Rolia J, Vetland V. Parameter estimation for performance models of distributed application systems. In: Proceedings of the 1995 Centre for Advanced Studies on Collaborative Research Conference. Toronto, CA: ACM, 1995. 54-63
[20]	Rolia J, Kalbasi A, Krishnamurthy D, Dawson S. Resource demand modeling for multi-tier services. In: Proceedings of the 1st joint WOSP/SIPEW International Conference on Performance Engineering. New York, USA: ACM, 2010. 207-216
[21]	Sun X. Estimating Resource Demands for Application Services [Master dissertation], Carleton University, Canada, 1999
[22]	Lu Y, Abdelzaher T, Lu C Y, Sha L, Liu X. Feedback control with queueing-theoretic prediction for relative delay guarantees in Web servers. In: Proceedings of the 2003 Real Time and Embedded Technology and Applications Symposium. Toronto, Canada: IEEE, 2003. 208-217
[23]	Kraft S, Pacheco-Sanchez S, Casale G, Dawson S. Estimating service resource consumption from response time measurements. In: Proceedings of the 2009 Performance Evaluation Methodologies and Tools. Coleraine, UK: ACM, 2009, 1-10
[24]	Menasce D. Computing missing service demand parameters for performance models. In: Proceedings of the 2008 Computer Measurement Group. Las Vegas, USA: CiteSeer, 2008. 241-248
[25]	Zheng T, Woodside M. Performance model estimation and tracking using optimal filters. IEEE Transactions on Software Engineering, 2008, 34(3): 391-406
[26]	Woodside M, Zhen T, Litoiu M. Service system resource management based on a tracked layered performance model. In: Proceedings of the 2006 International Conference on Autonomic Computing. Washington, USA: IEEE, 2006. 175-184