File Access Popularity Prediction for Hierarchical Storage for High-Energy Physics

doi:10.19678/j.issn.1000-3428.0057273

Computer Engineering ›› 2021, Vol. 47 ›› Issue (2): 126-132. doi: 10.19678/j.issn.1000-3428.0057273

• Advanced Computing and Data Processing • Previous Articles Next Articles

File Access Popularity Prediction for Hierarchical Storage for High-Energy Physics

CHENG Zhenjing^1,2, WANG Lu^1,2, CHENG Yaodong^1,2,3, CHEN Gang¹, HU Qingbao¹, LI Haibo^1,2

1. Institute of High Energy Physics, Chinese Academy of Sciences, Beijing 100049, China;
2. University of Chinese Academy of Sciences, Beijing 100049, China;
3. Tianfu Cosmic Ray Research Center, Institute of High Energy Physics, Chinese Academy of Sciences, Chengdu 610041, China

Received:2020-01-20 Revised:2020-02-28 Online:2021-02-15 Published:2020-03-25

面向高能物理分级存储的文件访问热度预测

程振京^1,2, 汪璐^1,2, 程耀东^1,2,3, 陈刚¹, 胡庆宝¹, 李海波^1,2

1. 中国科学院高能物理研究所, 北京 100049;
2. 中国科学院大学, 北京 100049;
3. 中国科学院高能物理研究所天府宇宙线研究中心, 成都 610041

作者简介:程振京(1993-),男,博士研究生,主研方向为分布式存储、机器学习;汪璐,副研究员、博士;程耀东、陈刚,研究员、博士;胡庆宝,硕士;李海波,副研究员、博士。
基金资助:
国家重点研发计划（2017YFB0203200）；国家自然科学基金（11675201，11805226，11805223）。

Abstract

Abstract: Computing for high-energy physics is typically data-intensive.It mainly adopts file-based hierarchical storage solutions where data is allocated based on the access popularity to storage devices with different performances.The existing schemes of data popularity prediction generally adopt a heuristic algorithm based on artificial experience,whose prediction accuracy is low.This paper proposes a method of predicting future access popularity using Long Short-Term Memory(LSTM) network,which consists of network structure design,training,and prediction algorithms.The method divides the dynamic time window to construct a time series of file access features,and on this basis predicts the access trends of different data.Experimental results on the data set of LHAASO high-energy physics experiments show that compared with SVM,MLP and other algorithms,the proposed method increases the prediction accuracy by about 30%,and it has stronger applicability.

Key words: hierarchical storage, file access characteristics, time series data, Long Short-Term Memory(LSTM) network, file access popularity

摘要： 高能物理计算是典型的数据密集型计算，其主要采用基于文件的分级存储方案，根据访问热度的不同将数据存储于不同性能的存储设备上，然而当前数据热度预测采用基于人工经验的启发式算法，准确率较低。提出一种借助长短期记忆网络预测文件未来访问热度的方法，包括网络结构设计、训练和预测算法等。该方法通过划分动态时间窗口构造文件访问特征的时序序列，预测不同数据的访问趋势。在LHAASO高能物理实验数据集上的实验结果表明，与SVM、MLP等算法相比，该方法预测准确率提升了30%左右，具有更强的适用性。

关键词: 分级存储, 文件访问特征, 时序数据, 长短期记忆网络, 文件访问热度

CLC Number:

TP391

CHENG Zhenjing, WANG Lu, CHENG Yaodong, CHEN Gang, HU Qingbao, LI Haibo. File Access Popularity Prediction for Hierarchical Storage for High-Energy Physics[J]. Computer Engineering, 2021, 47(2): 126-132.

程振京, 汪璐, 程耀东, 陈刚, 胡庆宝, 李海波. 面向高能物理分级存储的文件访问热度预测[J]. 计算机工程, 2021, 47(2): 126-132.

/ Recommend / Download Citations

URL: https://www.ecice06.com/EN/10.19678/j.issn.1000-3428.0057273

https://www.ecice06.com/EN/Y2021/V47/I2/126

Figures/Tables 8

References 20

[1]	CAO Zhen,CHEN Mingjun,CHEN Songzhan,et al.Introduction to large high altitude air shower observatory[J].Chinese Astronomy and Astrophysics,2019,43(4):457-478.
[2]	SCHWAN P.Lustre:building a file system for 1000-node clusters[EB/OL].[2019-12-10].http://www.clusterfs.com.
[3]	PETERS A J,SINDRILARU E A,ADDE G.EOS as the present and future solution for data storage at CERN[J].Journal of Physics,2015,664(4):42-62.
[4]	MEDDEB M,DHRAIEF A,BELGHITH A,et al.Least fresh first cache replacement policy for NDN-based IoT networks[J].Pervasive and Mobile Computing,2019,52:60-70.
[5]	LIU Aigui,CHEN Gang.A user-based LNS file prediction model[J].Computer Engineering and Applications,2007,43(29):14-16.(in Chinese)刘爱贵,陈刚.一种基于用户的LNS文件预测模型[J].计算机工程与应用,2007,43(29):14-16.
[6]	ZHANG G,CHIU L,DICKEY C,et al.Automated lookahead data migration in SSD-enabled multi-tiered storage systems[C]//Proceedings of IEEE Symposium on Mass Storage Systems & Technologies.Washington D.C.,USA:IEEE Press,2010:135-148.
[7]	SUN Ziheng,DI Liping,FANG Hui.Using long short-term memory recurrent neural network in land cover classification on landsat and cropland data layer time series[J].International Journal of Remote Sensing,2019,40(2):593-614.
[8]	GOODFELLOW I,BENGIO Y,COURVILLE A.Deep learning[M].[S.1.]:MIT Press,2016.
[9]	CHENG Gaofeng,LI Xin,YAN Yonghong.Using highway connections to enable deep small-footprint LSTM-RNNs for speech recognition[J].Chinese Journal of Electronics,2019,28(1):107-112.
[10]	FARHAN W,TALAFHA B,ABUAMMAR A,et al.Unsupervised dialectal neural machine translation[J].Information Processing and Management,2020,57(3):334-348.
[11]	KONG Weicong,DONG Zhaoyan,JIA Youwei,et al.Short-term residential load forecasting based on LSTM recurrent neural network[J].IEEE Transactions on Smart Grid,2017,10(1):841-851.
[12]	LIMA F D,AMARAL G M R,DE MOURA L G,et al.Predicting failures in hard drives with lSTM networks[C]//Proceedings of 2017 Brazilian Conference on Intelligent Systems.Washington D.C.,USA:IEEE Press,2017:222-227.
[13]	GREFF K,SRIVASTAVA R K,KOUTNIK J,et al.LSTM:a search space odyssey[J].IEEE Transactions on Neural Networks and Learning Systems,2016,28(10):2222-2232.
[14]	ABADI M,BARHAM P,CHEN J,et al.TensorFlow:a system for large-scale machine learning[C]//Proceedings of the 12th IEEE Symposium on Operating Systems Design and Implementation.Washington D.C.,USA:IEEE Press,2016:265-283.
[15]	GERON A.Hands-on machine learning with Scikit-learn,Keras,and TensorFlow:concepts,tools,and techniques to build intelligent systems[M].[S.1.]:O'Reilly Media,2019.
[16]	RUBIO G,POMARES H,ROJAS I,et al.A heuristic method for parameter selection in LS-SVM:application to time series prediction[J].International Journal of Forecasting,2011,27(3):725-739.
[17]	KINGMA D P,BA J.Adam:a method for stochastic optimization[EB/OL].[2019-12-10].https://arxiv.org/abs/1412.6980.
[18]	JOACHIMS T.Making large-scale SVM learning practical[M].Cambridge,USA:MIT Press,1998.
[19]	TANG Jiexiong,DENG Chenwen,HUANG Guangbin.Extreme learning machine for multilayer perceptron[J].IEEE Transactions on Neural Networks and Learning Systems,2015,27(4):809-821.
[20]	TSOU T S.A robust likelihood approach to inference about the kappa coefficient for correlated binary data[J].Statistical Methods in Medical Research,2019,28(4):1188-1202.

Please choose a citation manager

Content to export