基于用户行为的超级计算机作业失败预测方法

计算机工程与科学 ›› 2022, Vol. 44 ›› Issue (10): 1753-1761.

基于用户行为的超级计算机作业失败预测方法

唐阳坤1,2，鲜港1,2，杨文祥2,3，喻杰2，张晓蓉1，王耀彬1

(1.西南科技大学计算机科学与技术学院，四川绵阳 621010；
2.中国空气动力研究与发展中心计算空气动力研究所，四川绵阳 621050;3.国防科技大学计算机学院，湖南长沙 410073)

收稿日期:2021-09-02 修回日期:2022-01-10 接受日期:2022-10-25 出版日期:2022-10-25 发布日期:2022-10-28
基金资助:
国家自然科学基金（61872304,61802320）;空气动力学国家重点实验室基金（SKLA20200203）

Job failure prediction based on user behavior on supercomputers

TANG Yang-kun1,2,XIAN Gang1,2,YANG Wen-xiang2,3,YU Jie2,ZHANG Xiao-rong1,WANG Yao-bin1

(1.School of Computer Science and Technology,Southwest University of Science and Technology,Mianyang 621010;
2.Computational Aerodynamics Institute,China Aerodynamics Research and Development Center,Mianyang 621050;
3.College of Computer Science and Technology,National University of Defense Technology,Changsha 410073,China)

Received:2021-09-02 Revised:2022-01-10 Accepted:2022-10-25 Online:2022-10-25 Published:2022-10-28

摘要/Abstract

摘要： 超级计算机的规模不断扩大，与此同时，科学应用的复杂性也在不断增加，这导致了超级计算机上许多作业失败。作业失败会造成资源浪费，排队作业等待时间延长，严重影响系统的执行效率。提前预测作业失败，就可以采取必要的措施提升系统资源利用率和系统执行效率，这对未来的E级超级计算机至关重要。为此，尝试研究从已知的传统特征和构建特征中预测作业失败，发现能够反映用户工作行为模式和提交行为模式的特征及处理方式。通过结合行为特征和传统特征，提出基于树结构模型的综合框架来预测作业失败。实验结果表明，预测效果优于其他相关方法。

关键词: 系统执行效率, 作业日志分析, 用户行为, 作业失败预测, 机器学习

Abstract: The scale of supercomputers is expanding. Meanwhile, the complexity of scientific applications is also increasing, which leads to many job failures on supercomputers. These failed jobs causes a waste of resources and prolong the waiting time of queuing jobs, which seriously affects the reliability of the system. If these failed jobs can be predicted in advance, necessary measures can be taken to improve the system resource utilization and system execution efficiency, which is very important for the future exascale supercomputers. Therefore, this paper attempts to predict these job failures from the known traditional features and construction features, and find the features and processing methods that can reflect the users work behavior patterns and submission behavior patterns. By combining behavior features and traditional features, a comprehensive framework based on tree structure model is proposed to predict job failure. The prediction experimental results show that the comprehensive prediction framework is better than the single model prediction, and the comparative experimental results show that the prediction effect is better than other related methods.

Key words: system execution efficiency, job log analysis, user behavior, job failure prediction, machine learning

唐阳坤, 鲜港, 杨文祥, 喻杰, 张晓蓉, 王耀彬. 基于用户行为的超级计算机作业失败预测方法[J]. 计算机工程与科学, 2022, 44(10): 1753-1761.

TANG Yang-kun, XIAN Gang, YANG Wen-xiang, YU Jie, ZHANG Xiao-rong, WANG Yao-bin. Job failure prediction based on user behavior on supercomputers[J]. Computer Engineering & Science, 2022, 44(10): 1753-1761.

参考文献［17］

［1］	Yoo W,Sim A,Wu K.Machine learning based job status prediction in scientific clusters［C］∥Proc of 2016 SAI Computing Conference (SAI),2016:44-53.
［2］	Soualhia M,Khomh F,Tahar S.Predicting scheduling failures in the cloud: A case study with Google clusters and Hadoop on Amazon EMR［C］∥Proc of 2015 IEEE 17th International Conference on High Performance Computing and Communications,2015 IEEE 7th International Symposium on Cyberspace Safety and Security,and 2015 IEEE 12th International Conference on Embedded Software and Systems,2015: 58-65.
［3］	Grzonka D,Jakobik A,Kolodziej J,et al.Using a multi-agent system and artificial intelligence for monitoring and improv- ing the cloud performance and security［J］.Future Generation Computer Systems,2018,86: 1106-1117.
［4］	Chen X,Lu C D,Pattabiraman K.Failure analysis of jobs in compute clouds: A Google cluster case study［C］∥Proc of 2014 IEEE 25th International Symposium on Software Reliability Engineering,2014: 167-177.
［5］	Reiss C,Tumanov A,Ganger G R,et al.Heterogeneity and dynamicity of clouds at scale: Google trace analysis［C］∥Proc of the 3rd ACM Symposium on Cloud Computing,2012: 1-13.
［6］	Rosà A,Chen L Y,Birke R,et al.Demystifying casualties of evictions in big data priority scheduling［J］.ACM SIGMETRICS Performance Evaluation Review,2015,42(4): 12-21.
［7］	Rosa A,Chen L Y,Binder W.Understanding the dark side of big data clusters: An analysis beyond failures［C］∥Proc of the 45th Annual IEEE/IFIP International Conference on Dependable Systems and Networks,2015: 207-218.
［8］	Jayanthi R,Florence L.Software defect prediction techniques using metrics based on neural network classifier［J］.Cluster Computing,2019,22(1): 77-88.
［9］	Kumaresan K,Ganeshkumar P.Software reliability modeling using increased failure interval with ANN［J］.Cluster Computing,2019,22(2): 3095-3102.
［10］	Padhy N,Singh R P,Satapathy S C.Cost-effective and fault-resilient reusability prediction model by using adaptive genetic algorithm based neural network for web-of-service applications［J］.Cluster Computing,2019,22(6): 14559-14581.
［11］	Manjula C, Florence L.Deep neural network based hybrid approach for software defect prediction using software metrics［J］.Cluster Computing,2019,22(4): 9847-9863.
［12］	Hamerly G,Elkan C.Bayesian approaches to failure prediction for disk drives［C］∥Proc of International Conference on Machine Learning,2001:202-209.
［13］	Nie B,Xue J,Gupta S,et al.Machine learning models for GPU error prediction in a large scale HPC system［C］∥Proc of the 48th Annual IEEE/IFIP International Conference on Dependable Systems and Networks,2018: 95-106.
［14］	Das A,Mueller F,Siegel C,et al.Desh: Deep learning for system health prediction of lead times to failure in HPC［C］∥Proc of the 27th International Symposium on High- Performance Parallel and Distributed Computing,2018: 40-51.
［15］	Schroeder B,Gibson G A.A large-scale study of failures in high-performance computing systems［J］.IEEE Transactions on Dependable and Secure Computing,2009,7(4): 337-350.
［16］	Snir M,Wisniewski R W,Abraham J A,et al.Addressing failures in exascale computing［J］.The International Journal of High Performance Computing Applications,2014,28(2): 129-173.
［17］	Liu Chun-hong,Han Jing-jing,Shang Yan-lei.Predicting job failure in cloud cluster: Based on SVM classification［J］.Journal of Beijing University of Posts and Telecommunications,2016,39(5):104-109.(in Chinese)
［18］	Nakka N,Agrawal A,Choudhary A.Predicting node failure in high performance computing systems from failure and usage logs［C］∥Proc of 2011 IEEE International Symposium on Parallel and Distributed Processing Workshops and PhD Forum,2011: 1557-1566.
［19］	Georgiou Y,Cadeau T,Glesser D,et al.Energy accounting and control with SLURM resource and job management system［C］∥Proc of International Conference on Distributed Computing and Networking,2014: 96-118.
［20］	Bergroth L,Hakonen H,Raita T.A survey of longest common subsequence algorithms［C］∥Proc of the 7th International Symposium on String Processing and Information Retrieval,2000: 39-48.
［21］	Likas A,Vlassis N,Verbeek J J.The global k-means cluster- ing algorithm［J］.Pattern Recognition,2003,36(2): 451-461.
［22］	Benesty J,Chen J,Huang Y,et al.Pearson correlation coefficient［M］∥Noise Reduction in Speech Processing.Berlin,Heidelberg:Springer,2009: 1-4.
［23］	Liaw A,Wiener M.Classification and regression by randomForest［J］.R News,2002,2/3:18-22.
［24］	Abhishek L.Optical character recognition using ensemble of SVM,MLP and extra trees classifier［C］∥Proc of 2020 International Conference for Emerging Technology,2020:1-4.
［25］	Rosa A,Chen L Y,Binder W.Predicting and mitigating jobs failures in big data clusters［C］∥Proc of the 15th IEEE/ACM International Symposium on Cluster,Cloud and Grid Computing,2015: 221-230.
	附中文参考文献:
［17］	刘春红,韩晶晶,商彦磊.基于 SVM 分类的云集群失败作业主动预测方法［J］.北京邮电大学学报,2016,39(5):104-109.

[1]	赵振宇, 杨天豪, 蒋汶乘, 张书政. 基于机器学习的多压多温多参标准单元延迟快速计算方法[J]. 计算机工程与科学, 2023, 45(08): 1331-1338.
[2]	李小玲, 方建滨, 马俊, 谭霜, 谭郁松. 基于监督学习的稀疏矩阵自动任务分配[J]. 计算机工程与科学, 2023, 45(05): 782-789.
[3]	刘云, 肖添, 王梓宇. 动态特征选择算法对恶意行为检测的优化研究[J]. 计算机工程与科学, 2022, 44(04): 665-673.
[4]	李文丽. 基于朴素贝叶斯分类的网络谣言识别研究[J]. 计算机工程与科学, 2022, 44(03): 495-501.
[5]	马满福1,2，员欣淼1,2，李勇1,2，刘元喆1,2，王常青3. 虚拟空间中社会分层行为研究[J]. 计算机工程与科学, 2020, 42(05): 803-811.
[6]	欧嵬1，王勇军1，韩文报2. 基于用户行为的可信模型研究[J]. J4, 2013, 35(5): 46-50.
[7]	王振宇1,郭力 2. 基于Hadoop的搜索引擎用户行为分析[J]. J4, 2011, 33(4): 115-120.
[8]	付关友朱征宇. 个性化服务中基于行为分析的用户兴趣建模[J]. J4, 2005, 27(12): 76-78.