具负载平衡之IPTV影音串流系统

2014-12-18 18:04:54郭书耀李孟晃
现代电子技术 2014年24期

郭书耀+李孟晃

摘  要: 影音媒体具高带宽及实时特性,因此影音串流系统的服务质量的提升端赖影音服务器的有效管理。在此提出一套Video replication及Access request distribution的算法,藉由将热门影音分散到各影音服务器以及影音存取的分散,以达到各影音服务器系统负载平衡的效果。在此比较了3个不同的算法的负载平衡的状况,经实例验证,提出的算法确能达到影音服务器负载平衡的效果,承载更多的系统服务要求。该系统构建在Smart?iTV的IPTV系统台,并讨论了实作的相关技术。

关键词: IPTV; 负载平衡; 影音串流; 影音备份; 随选视讯

中图分类号: TN964?34                   文献标识码: A                            文章编号: 1004?373X(2014)24?0022?06

IPTV system with load balance among replica servers

GUO Shu?yao, LI Meng?huang

(Department of Information Technology and management, Shih Chien University, Taipei 104, Taiwan, China)

Abstract: Since the video media has wide band width, the improvement of QoS (quality of service) of IPTV system depends on effective management of video servers. A video replication and access request distribution algorithm to implement a load balanced IPTV system with replica servers is proposed in this paper. The proposed algorithm is based on the more replicas for hot program and load distribution among replica servers. By comparing with two other algorithms, the proposed algorithm features better load balance effects. System practices demonstrate that the access requests can be evenly distributed among replica servers such that a better streaming service can be provided. The proposed algorithm has been implemented on an IPTV system platform. The relative technical issues for the implementation are discussed.

Keyword: IPTV; load balance; video streaming; video replica; video on demand

0  引  言

在VOD(Video on Demand)的系统规划上或是Video Replication and Placement的研究上,或是CDN(Content Delivery Network)系统建置的成本问题探讨上,storage capacity的需求是很明确的,就是各Video program的档案总和(也就是各Video program的program duration总和)。然而network bandwidth的需求的估算就比较困难[1?3]。论文以“worst?case demand”作为network bandwidth配置的参考参数,从而决定系统建置所需的storage capacity、网络光纤及交换设备的容量。Worst?case demand即是peak usage demand,在peak usage period的时候,使用者的数目达到最高峰,这些使用者所使用的network bandwidth总和即是worst?case demand,系统网络的配置必须依据此最高峰的network bandwidth需求来设定,以确保其服务质量。而在文献[1?3]的论文中,则以replication的方式来分散单一Origin server的负荷,在此架构中,Replica servers用来储存许多热门的影音内容,各影音内容依其viewing request probability来决定影音replication的份数,从而分散这些热门影音内容被点选时的网络流量带宽,以得到最佳成本的系统建置。 虽然上述论文有数学模式来描述worst?case demand的估量方法,然而还是无法精确地估量系统的worst?case demand,尤其Replica servers对worst?case demand估量的影响,文献[1?3]的分析模式有其适用性的问题与困难。……

登录APP查看全文