基于时序相关性的云平台多负载序列联合预测

doi:10.13190/j.jbupt.2019-244

北京邮电大学学报 ›› 2020, Vol. 43 ›› Issue (4): 68-75.doi: 10.13190/j.jbupt.2019-244

基于时序相关性的云平台多负载序列联合预测

张志华¹, 王梦情¹, 毛文涛^1,2, 刘春红^1,2, 程渤³

1. 河南师范大学计算机与信息工程学院, 新乡 453007;
2. 河南师范大学智慧商务与物联网技术河南省工程实验室, 新乡 453007;
3. 北京邮电大学网络与交换技术国家重点实验室, 北京 100876

收稿日期:2019-11-18 发布日期:2020-08-15
通讯作者: 刘春红(1969-),女,副教授,E-mail:lch@htu.edu.cn. E-mail:lch@htu.edu.cn
作者简介:张志华(1994-),女,硕士生.
基金资助:
国家自然科学基金项目（U1704158）；河南省重点研发与推广专项（科技攻关）项目（202102210163）；河南师范大学博士启动基金项目（5101119170145）；河南省高等教育教学改革研究与实践项目（2019SJGLX033Y）

Joint Prediction of Multi-Workload Sequences Based on Temporal Correlation in the Cloud

ZHANG Zhi-hua¹, WANG Meng-qing¹, MAO Wen-tao^1,2, LIU Chun-hong^1,2, CHENG Bo³

1. School of Computer and Information Engineering, Henan Normal University, Xinxiang 453007, China;
2. Engineering Laboratory of Intelligence Business & Internet of Things, Henan Normal University, Xinxiang 453007, China;
3. State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China

Received:2019-11-18 Published:2020-08-15

摘要/Abstract

摘要： 为提高云平台负载预测的精度，提出了一种基于时序相关性的多负载序列联合预测方法.首先，为获得相似的负载序列，采用长短时记忆网络提取负载序列的时序特征，再利用层次聚类法，获得在时序特征空间相似的负载序列类；其次，对获得的每个负载序列类分别构建多任务学习模型，挖掘和利用负载序列间隐藏的共享领域知识，提高模型泛化能力和预测精度，并实现多个负载序列的联合预测.使用Google数据集的中央处理器负载监控日志进行验证，结果表明，时序特征聚类可有效提取和利用负载序列的全局时序特征，降低原始序列的噪声，获得特征上相似的序列；与常用的负载预测方法比，所提方法对不同变化规律的负载序列都具有更精确的预测效果.

关键词: 云计算, 负载, 时序特征, 聚类, 结构化预测

Abstract: A novel approach,joint prediction of multi-workload sequences,was proposed based on temporal correlation. Firstly,long short-term memory was used to extract the temporal feature among workload sequences for obtaining the similar workload sequences,while hierarchical clustering algorithm was used to obtain workload sequences clusters. Then,construct multi-task learning model respectively for each obtained sequence clusters,capture and utilize the shared domain knowledge among multiple workload sequences,so as to achieve joint prediction of multiple workload sequences as well as improve generalization ability and prediction accuracy of the model. Results of experiment on dataset of Google cluster trace 2011 demonstrates that the temporal feature clustering can effectively extract and utilize the global temporal feature of workload sequences,reduce the noise of original sequences and get workload sequences clusters with similar characteristics. Proposed method performs better in prediction accuracy than the state-of-the-art methods.

Key words: cloud computing, workload, temporal feature, cluster, structural prediction

中图分类号:

TP393

张志华, 王梦情, 毛文涛, 刘春红, 程渤. 基于时序相关性的云平台多负载序列联合预测[J]. 北京邮电大学学报, 2020, 43(4): 68-75.

ZHANG Zhi-hua, WANG Meng-qing, MAO Wen-tao, LIU Chun-hong, CHENG Bo. Joint Prediction of Multi-Workload Sequences Based on Temporal Correlation in the Cloud[J]. Journal of Beijing University of Posts and Telecommunications, 2020, 43(4): 68-75.

参考文献

[1] Al-Dhuraibi Y, Paraiso F, Djarallah N, et al. Elasticity in cloud computing:state of the art and research challenges[J]. IEEE Transactions on Services Computing, 2017, 11(2):430-447.
[2] Yang Jingqi, Liu Chuanchang, Shang Yanlei, et al. A cost-aware auto-scaling approach using the workload prediction in service clouds[J]. Information Systems Frontiers, 2014, 16(1):7-18.
[3] Calheiros R N, Masoumi E, Ranjan R, et al. Workload prediction using ARIMA model and its impact on cloud applications' QoS[J]. IEEE Transactions on Cloud Computing, 2014, 3(4):449-458.
[4] Liu Chunhong, Liu Chuanchang, Shang Yanlei, et al. An adaptive prediction approach based on workload pattern discrimination in the cloud[J]. Journal of Network and Computer Applications, 2017, 80:35-44.
[5] Islam S, Keung J, Lee K, et al. Empirical prediction models for adaptive resource provisioning in the cloud[J]. Future Generation Computer Systems, 2012, 28(1):155-162.
[6] Hoang M N, Gaurav K, Daeyoung K. Host load prediction in cloud computing using long short-term memory encoder-decoder[J]. The Journal of Supercomputing, 2019, 75(11):7592-7605.
[7] Verma A, Pedrosa L, Korupolu M, et al. Large-scale cluster management at Google with Borg[C]//ECCS 2015:Proceedings of the Tenth European Conference on Computer Systems. New York:ACM, 2015:18-35.
[8] Jiang Yexi, Perng Changshing, Li Tao, et al. Cloud analytics for capacity planning and instant vm provisioning[J]. IEEE Transactions on Network and Service Management, 2013, 10(3):312-325.
[9] Parminder S, Pooja G, Kiran J. Tasm:technocrat arima and SVR model for workload prediction of web applications in cloud[J]. Cluster Computing, 2019, 22(2):619-633.
[10] Kumar J, Goomer R, Singh A K. Long short term memory recurrent neural network (LSTM-RNN) based workload forecasting model for cloud datacenters[J]. Procedia Computer Science, 2018, 125:676-682.
[11] Janardhanan D, Barrett E. CPU workload forecasting of machines in data centers using LSTM recurrent neural networks and ARIMA models[C]//ICITST 2017:the 12th International Conference for Internet Technology and Secured Transactions. Cambridge:IEEE, 2017:55-60.
[12] Zhou Jiayu, Chen Jianhui, Ye Jieping. Malsar:multi-task learning via structural regularization[EB/OL]. 2012(2012-12-18)[2019-11-10]. http://www.MALSAR.org.
[13] 徐前方, 王嘉春, 肖波. 融合时空上下文信息的兴趣点推荐[J]. 北京邮电大学学报, 2018, 41(1):37-42, 50. Xu Qianfang, Wang Jiachun, Xiao Bo. Point-of-interest recommendation with spatio-temporal context awareness[J]. Journal of Beijing University of Posts and Telecommunications, 2018, 41(1):37-42, 50.
[14] Reiss C, Wilkes J, Hellerstein J L. Google cluster-usage traces:format+ schema[EB/OL]. 2014(2014-11-17)[2019-11-10]. https://github.com/google/cluster-data.
[15] Han Li, Romero C E, Yao Zheng. Wind power forecasting based on principle component phase space reconstruction[J]. Renewable Energy, 2015, 81:737-744.

基于时序相关性的云平台多负载序列联合预测

Joint Prediction of Multi-Workload Sequences Based on Temporal Correlation in the Cloud

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	朱瑞文王雅文林欢李帅赟吴彦芳. 一种基于层次聚类的测试用例集约简方法[J]. 北京邮电大学学报, 2023, 46(4): 9-14.
[2]	张小乾, 王潇, 薛旭倩, 谈振, 蒲磊. 基于加权多核子空间聚类的图像分割方法[J]. 北京邮电大学学报, 2023, 46(3): 78-83.
[3]	黄跃珍戴晶帼张承业魏东. 基于关联图关键边发现的人脸图像聚类算法[J]. 北京邮电大学学报, 2023, 46(1): 97-102.
[4]	苏扬, 魏连锁, 郭媛. 多目标融合势博弈无线自组网络拓扑控制算法[J]. 北京邮电大学学报, 2022, 45(4): 91-97.
[5]	苏扬魏连锁郭媛. 多目标融合的序数势博弈无线自组网络拓扑控制算法[J]. 北京邮电大学学报, 2022, 45(4): 105-111.
[6]	王永, 尹恩民, 冉珣. 基于BC聚类的差分隐私保护推荐算法[J]. 北京邮电大学学报, 2021, 44(2): 81-88.
[7]	温卫. 基于密度聚类的容迟网络路由协议[J]. 北京邮电大学学报, 2020, 43(5): 137-142.
[8]	何倩, 李双富, 黄焕, 徐红. 一种海量数据快速聚类算法[J]. 北京邮电大学学报, 2020, 43(3): 118-124.
[9]	阎逸飞, 王智立, 邱雪松, 王嘉潞. Spark环境下基于数据倾斜模型的Shuffle分区优化方案[J]. 北京邮电大学学报, 2020, 43(2): 116-121.
[10]	刘保菊, 喻鹏, 丰雷, 邱雪松, 江昊. 电力SDN通信网中面向负载均衡的路由重构[J]. 北京邮电大学学报, 2020, 43(2): 16-22.
[11]	马庆刘, 喻鹏, 吴佳慧, 熊翱, 颜拥. 基于深度强化学习的综合能源业务通道优化机制[J]. 北京邮电大学学报, 2020, 43(2): 87-93.
[12]	王璇, 侯蓉晖, 徐伟琳. 低轨卫星网络动态路径切换技术[J]. 北京邮电大学学报, 2020, 43(2): 80-86,109.
[13]	双锴, 李怡雯, 吕志恒, 韩静, 刘建伟. 基于归一化特征判别的日志模板挖掘算法[J]. 北京邮电大学学报, 2020, 43(1): 68-73.
[14]	苏东, 余宁梅. 基于深度学习的OBD端口占用状态自动识别算法[J]. 北京邮电大学学报, 2019, 42(6): 49-57.
[15]	张嘉驰, 刘留, 周涛, 王凯, 朴哲岩. 基于机器学习的无线信道簇的提取与轨迹追踪[J]. 北京邮电大学学报, 2019, 42(4): 126-132.