【作者】 李军； 彭耀； 温明杰； 任德虎； 赵致杰； 张陈斌； 陈宗海；

【Author】 Li Jun;Peng Yao;Wen Ming-Jie;Ren De-Hu;Zhao Zhi-Jie;Zhang Chen-Bing;Chen Zong-Hai;USTC-ETHINK Big Data Business Intelligence Joint Laboratory;Anhui ETHINK Information Technology Co,LTD;Department of Automation,University of Science and Technology of China;

【机构】 中国科大-象形大数据商业智能联合实验室； 安徽象形信息科技有限公司； 中国科学技术大学自动化系；

【摘要】 随着云计算与分布式集群技术的发展,大数据概念在数据存储、数据分析等多方面都有了更广的扩展和延伸,其中基于Spark的高性能计算技术近年来也得到了前所未有的重视。本文主要针对当代仿真工具无法对海量数据进行存储、计算等问题,提出了一个基于Spark高性能计算的新型仿真云平台设计,该平台不仅能够兼容集群利用现有计算资源灵活高效地处理海量数据,而且还具有良好的可扩展性,能够满足仿真领域机器学习场景的需求。更多还原

【Abstract】 With the development of cloud computing and distributed clustering technology,big data concepts in data storage,data analysis,and many have a wider expansion and extension,which are based on high-performance computing technology Spark in recent years has been unprecedented attention.This article can not mass data storage for modern simulation tools,computing and other issues,propose a new simulation platform based on cloud Spark high-performance computing,the platform is not only compatible with cluster computing using existing resources flexibly and efficiently handle massive amounts of data,but also has good scalability to meet the needs of the simulation in the field of machine learning scenarios.更多还原