摘要
随着在线教育热度的持续上升,常态化录播系统获得越来越多的关注。在大规模视频场景下,传统常态录播系统面临着可扩展性差、容错性差、视频检索以及分享困难的问题。设计实现了一套超融合架构常态录播系统,利用分布式对象存储的高可扩展性和高可靠性的特点,实现视频的稳定采集、可靠存储以及安全便捷的访问。实验结果表明,3台通用x86服务器组成的超融合集群,在满足读写时延需求的情况下,可以支撑600路2 Mbit/s码流视频并发写和4000路2 Mbit/s码流并发读。所设计的超融合架构录播系统目前已经应用在西安交通大学智能教室建设项目中,实现394间智能教室的常态化录播。
As the popularity of online education,normalization recording and broadcasting system is gaining increasing attention.In large-scale video scenario,traditional normalization recording and broadcasting system faces the challenge of reliability,extendibility,fault-tolerance,video index and sharing.A novel hyper-converged normalization recording and broadcasting system based on distributed object storage system is designed and realized.This system exploits high extensibility and reliability of distributed object storage system to satisfy large-scale video capturing,storage and sharing.The experimental results show that the hyper-converged cluster which composes of 3 general x86 servers is able to sustain 600 ways of 2 Mbit/s concurrency writing and 4000 ways of 2 Mbit/s concurrency reading.The system has been applied to Xi'an Jiaotong University smart class construction for 394 classrooms.
作者
王强
覃遵颖
崔靖茹
折波
李国栋
WANG Qiang;QIN Zunying;CUI Jingru;ZHE Bo;LI Guodong(Network Information Center,Xi'an Jiaotong University,Xi'an 710049,China)
出处
《通信学报》
EI
CSCD
北大核心
2018年第A01期220-226,共7页
Journal on Communications
基金
赛尔网络下一代互联网技术创新项目(No.NGII20160508)~~
关键词
常态录播
分布式对象存储
超融合
normalization recording system
distributed object storage
hyper-converged
