BT类型P2P_VOD的服务器可扩展性研究
|
摘要
随着Internet的迅速发展和宽带接入技术的日益普及,视频点播系统(VOD)已成为互联网上最流行的视频应用之一。如何在Internet上提供大规模的VOD服务成为现阶段亟待解决的问题,降低服务器端负载是提供大规模VOD服务的有效途径。为了满足视频点播系统高流畅度、高可扩展性及低播放延迟的需求,根据数据分发策略的不同,流媒体系统经历了以下几个阶段：(1)传统的客户机/服务器模式。在传统的C/S模式中,随着网络规模的增大,流媒体服务器很容易成为系统的瓶颈。(2)IP组播技术。由于组播管理、组播拥塞控制等问题导致IP组播技术无法在Internet上广泛应用。(3)CDN (Content Delivery Network)由于其部署、维护费用较高而难以大范围应用。
为了解决以上问题,P2P技术被广泛应用到流媒体系统中。在P2P流媒体系统中用户接入带宽不一致的现象普遍存在；另外,用户节点可以随时加入退出系统致使网络有很大的动态性；再加上流媒体本身对带宽占用率较高、服务时间长、有严格的播放次序及数据传输延迟要求,给P2P流媒体点播系统的研究带来了很大的挑战。
本文围绕Internet上难以大规模部署VOD服务这一背景,对基于BT的P2P流媒体点播系统进行了深入的研究,并给出了相应的解决方法。
将P2P技术应用到流媒体或者其他内容分发应用中的主要目标是充分利用终端用户的上传能力来提高流媒体应用的性能。提高流媒体系统中对等节点之间的数据请求命中率是降低媒体服务器的负载一个有效解决方法。首先,本文对文件共享模型进行扩展,使其适合于分析点播系统服务器负载。然后基于此扩展模型分析可共享数据片段数量及相互合作的对等节点数量对数据请求命中率的影响,通过提高对等节点数据请求命中率降低流媒体服务器负载。MATLAB实验结果表明,合理设置可共享数据块数量及协作节点集合大小可以降低服务器负载。
数据调度策略是流媒体点播系统的一个核心问题,好的数据调度策略可以使对等节点充分利用邻居节点的带宽资源及缓存数据块,减少访问流媒体服务器的次数,提高系统的可扩展性。本文提出了一种基于带宽的多源数据调度策略(BBMDS, Base-on-Bandwidth Multi-supplier Data Scheduling policy)。在数据调度过程中,充分考虑节点的异构性。根据数据块的紧急度和稀有度、邻居节点的缓存映像及带宽信息,在保证请求数据块在播放期限到达的前提下,达到充分利用邻居节点的带宽资源降低服务器负载的目标。从OMNET++仿真结果来看,BBMDS能有效降低服务器负载。
With the development of the Internet and the widely usage of the broadband access technology, VOD has become one of the most popular streaming applications over the Internet. Providing large-scale VOD services over the Internet has become an urgent problem needed be solved at the present stage, and reducing server load is an effective way. To meet the high flow, high scalability and low delay streaming requirement, steaming media system has gone through several stages according to the different data distribution strategies. (1) The traditional client/server mode. In this mode, as the increasing of size of network, the streaming media server could easily become the bottleneck of system. (2) IP multicast technology. It can't be widely used in the Internet due to the management and congestion control problem of multicast. (3) CDN (Content Delivery Network).Owing to its hard deployment and high cost, it is difficult to be widely applied.
In order to solve the problems above, P2P technology has been widely applied to the streaming media system. In the P2P streaming system, the problem of inconsistence of access bandwidth of users is widespread. In addition, the user mode can join and exit the system at any time, which results in very dynamic networks. What's more, high bandwidth, long service time, strict sequences of playing and transmission delay requirements have been bringing great challenge to P2P streaming media on-demang system.
This paper focuses on the difficulty of large-scale deployment of internet VOD service, and proposes the corresponding solution based on making future research on P2P streaming media system based on BT.
The main purpose to apply P2P technology to streaming media content is to make full use of the upload ability of end-users to improve streaming media applications performance. An effective solution is to improve the data requests hit rates between the peers to reduce server load. First, we expand on the file sharing model, making it suitable for analysis of on-demand system server load. Then, based on this model, analyzing the impact of the effective of data requests caused by fragment available for sharing and peer group size. The results of experiment on MATLAB show that a proper number of data blocks and collaboration size of collaboration nodes can reduce server load.
Data scheduling strategy is the core issue of the Streaming media system. A proper data scheduling strategy can take advantage of the peer's neighbors'bandwidth resources and cache data block, reduce the number of system access and improve system scalability. This paper presents a Bandwidth-based Multi-supplier Data Scheduling strategy, which take the heterogeneity of nodes into account fully in the data scheduling process. According to the emergency degree and rarity of the block, cache image and bandwidth of the neighbor nodes, ensuring that the requested data block can arrive at the premise time, take full advantage of the bandwidth resources of neighbors to reduce server load. OMNET++ simulation results show that BBNDS can reduce the server load effectively.
为了解决以上问题,P2P技术被广泛应用到流媒体系统中。在P2P流媒体系统中用户接入带宽不一致的现象普遍存在；另外,用户节点可以随时加入退出系统致使网络有很大的动态性；再加上流媒体本身对带宽占用率较高、服务时间长、有严格的播放次序及数据传输延迟要求,给P2P流媒体点播系统的研究带来了很大的挑战。
本文围绕Internet上难以大规模部署VOD服务这一背景,对基于BT的P2P流媒体点播系统进行了深入的研究,并给出了相应的解决方法。
将P2P技术应用到流媒体或者其他内容分发应用中的主要目标是充分利用终端用户的上传能力来提高流媒体应用的性能。提高流媒体系统中对等节点之间的数据请求命中率是降低媒体服务器的负载一个有效解决方法。首先,本文对文件共享模型进行扩展,使其适合于分析点播系统服务器负载。然后基于此扩展模型分析可共享数据片段数量及相互合作的对等节点数量对数据请求命中率的影响,通过提高对等节点数据请求命中率降低流媒体服务器负载。MATLAB实验结果表明,合理设置可共享数据块数量及协作节点集合大小可以降低服务器负载。
数据调度策略是流媒体点播系统的一个核心问题,好的数据调度策略可以使对等节点充分利用邻居节点的带宽资源及缓存数据块,减少访问流媒体服务器的次数,提高系统的可扩展性。本文提出了一种基于带宽的多源数据调度策略(BBMDS, Base-on-Bandwidth Multi-supplier Data Scheduling policy)。在数据调度过程中,充分考虑节点的异构性。根据数据块的紧急度和稀有度、邻居节点的缓存映像及带宽信息,在保证请求数据块在播放期限到达的前提下,达到充分利用邻居节点的带宽资源降低服务器负载的目标。从OMNET++仿真结果来看,BBMDS能有效降低服务器负载。
With the development of the Internet and the widely usage of the broadband access technology, VOD has become one of the most popular streaming applications over the Internet. Providing large-scale VOD services over the Internet has become an urgent problem needed be solved at the present stage, and reducing server load is an effective way. To meet the high flow, high scalability and low delay streaming requirement, steaming media system has gone through several stages according to the different data distribution strategies. (1) The traditional client/server mode. In this mode, as the increasing of size of network, the streaming media server could easily become the bottleneck of system. (2) IP multicast technology. It can't be widely used in the Internet due to the management and congestion control problem of multicast. (3) CDN (Content Delivery Network).Owing to its hard deployment and high cost, it is difficult to be widely applied.
In order to solve the problems above, P2P technology has been widely applied to the streaming media system. In the P2P streaming system, the problem of inconsistence of access bandwidth of users is widespread. In addition, the user mode can join and exit the system at any time, which results in very dynamic networks. What's more, high bandwidth, long service time, strict sequences of playing and transmission delay requirements have been bringing great challenge to P2P streaming media on-demang system.
This paper focuses on the difficulty of large-scale deployment of internet VOD service, and proposes the corresponding solution based on making future research on P2P streaming media system based on BT.
The main purpose to apply P2P technology to streaming media content is to make full use of the upload ability of end-users to improve streaming media applications performance. An effective solution is to improve the data requests hit rates between the peers to reduce server load. First, we expand on the file sharing model, making it suitable for analysis of on-demand system server load. Then, based on this model, analyzing the impact of the effective of data requests caused by fragment available for sharing and peer group size. The results of experiment on MATLAB show that a proper number of data blocks and collaboration size of collaboration nodes can reduce server load.
Data scheduling strategy is the core issue of the Streaming media system. A proper data scheduling strategy can take advantage of the peer's neighbors'bandwidth resources and cache data block, reduce the number of system access and improve system scalability. This paper presents a Bandwidth-based Multi-supplier Data Scheduling strategy, which take the heterogeneity of nodes into account fully in the data scheduling process. According to the emergency degree and rarity of the block, cache image and bandwidth of the neighbor nodes, ensuring that the requested data block can arrive at the premise time, take full advantage of the bandwidth resources of neighbors to reduce server load. OMNET++ simulation results show that BBNDS can reduce the server load effectively.
引文
[1]朱承,刘忠,张维明,等.结构化P2P网络中基于流言传播的负载均衡[J].通信学报,2004,25(4)：31-40.
[2]Tomoya K, Shigeki Y. Application of P2P (peer-to-peer) technology to marketing[C] Cyberworlds, Proceedings 2003 International Conference,2003:372-379.
[3]K P, Gummadi S, Saroiu S D. Gribble. A Measurement Study of Peer-to-Peer File Sharing Systems. Proceedings of Multimedia Computing and Networking, January 2002(MMCN' 02), San Jose, CA, USA.
[4]Clip2. Gnutella measurement project [M/OL].2001,5. http://www.clip2.com/.
[5]Nathaniel L, Matei R, Adam W. Wierzbicki. Deconstructing the Kazaa Network,3rd IEEE Workshop on Internet Applications (WIAPP'03) [C], Santa Clara, CA, USA,2003, 6:112-120.
[6]Chu Y H, Sanjay G, Rao S S. A Case for End System Multicast [J]. IEEE Journel on Selected Areas in Communications.2002,20(8):p.1456.
[7]Banerjee S, Bhattacharjee B, kommareddy C. Scalable application layer multicast[C]. Proceedings of the ACM SIGCOMM 2002, Pittsburgh, PA.2002,8.
[8]Deshpande H, Bawa M. Streaming Live Media over a Peer-to-peer Network [R].2001, Technical report, Stanford University.
[9]Deshpande H, Bawa M, Garacia H M. Streaming live media over peers [R]. Tech. Rep, Stanford University,2002,3.
[10]Guo Y, Suh K, Kurose J, et al. P2cast:Peer-to-peer patching scheme for VoD service [A]. In Proc. of the 12th Int'1 Conf. on World Wide Web[C], New York:ACM Press, 2003:301-309.
[11]Eager D, Vernon M, Zahorjan J. Minimizing bandwidth requirements for on-demand data delivery [J]. IEEE Trans. On Knowledge and Data Engineering,2001,13(5): 742-757.
[12]Padmanabhan V, Wang J, Chou A, et al. Distributing streaming media content using cooperative networking[C]. Proceedings of ACM/IEEE NOSSDAV. Miami, FL, USA.2002.
[13]Castro M, Druschel P, Kermarrec A, et al. SplitStream:High-bandwidth multicast in a cooperative environment [J]//Proceedings of 19th ACM Sysposium on operating System Principles. Lake Bolton, New York.2003,3:292-303.
[14]Rowstron A, Druschel P. Pastry:Scalable, Decentralized Object Location and Routing for Large-Scale Peer-to-Peer System [J]//Computer Science,2001(2218): 329-350.
[15]Zhang X, Liu J, Li B, et al. CoolStreaming/DONet:A data-driven overlay network for efficient live media streaming. [C]//Proceedings of IEEE INFOCOM,2005,3: 2102-2111.
[16]杨天路,刘宇宏,张文,等.P2P网络技术原理与系统开发案例[M].北京：人民邮电出版社,2007.
[17]Shi S, Turner J. Multicast routing and bandwidth dimensioning in overlay networks [J]. IEEE Journal on Selected Areas in Communications,2002,20(8):1444-1455.
[18]Do T, Hua T, Kien A. P2VoD:providing fault tolerant video-on-demand streaming in peer-to-peer environment[C]//Communications,2004 IEEE International Conference,2004,3:1467-1472.
[19]Gang W, Tzi-cker C. How efficient is BitTorrent? [C]//Multimedia Computing and Networking, San Jose, CA, USA,2006.
[20]Aguilar A, Ege R K, Makki K, et al. Enabling Peer Cooperation in Private Local Area Networks using BitTorrent[C]//Computers and Communications.2007,615-622.
[21]Mario B, Alfio L. An Analytical Model of a BitTorrent Peer [C]//Distributed and Network-Based Processing,2007:482-489.
[22]Comparison between the different BitTorrent tracker modes [EB/OL]. http://www.phpsoft.org/trackermodes.php.
[23]Thommes R, Coates M. BitTorrent Fairness:Analysis and Improvements[R]. Department of Electrical and Computer Engineering,2004,5.
[24]蒋雪岭.P2P流媒体系统多点下载技术的实现研究[J].沈阳大学学报,2006,18(4)：60-63.
[25]Bram Cohen. Incentives Build Robustness in BitTorrent. Workshop on economics of P2P system.2003,6.
[26]张杰,张尧,黄道颖,等.P2P网络中一种改进的片段选择控制策略[J].微计算机信息,2007,23(21)：140-142.
[27]Ying L, Yang L. Experimental Views on Neighbor Selection in BitTorrent[C]. Network and Parallel Computing Workshops NPC Workshops.2007:813-818.
[28]刘永卫,王东.BitTorrent系统中一种自适应阻塞算法[J].计算机工程与用,2007,43(34)：122-124.
[29]Dana C, Li D, Harrison D, et al. BASS:BitTorrent assisted streaming system for video-on-demand[C]//Multimedia Signal Processing,2005 IEEE 7th Workshop on Shanghai,2005:1-4.
[30]Vlavianos A, Iliofotou M, Faloutsos M. BiToS:Enhancing BitTorrent for Supporting Streaming Applications [C]//Computer Communications Processing, INFOCOM 2006. 25th IEEE International conference, Barcelona,2006:1-6.
[31]Qiu D, Srikant R. Modeling and Performance Analysis of BitTorrent-like peer-to-Peer Networks. Proc[C]. ACM SIGCOMM, Karlsruhe, Germany,2004,8: 367-378.
[32]郑伟平,齐德昱,徐克付.BitTorrent类型P2P流媒体系统研究[J].计算机科学,2010,37(2)：123-125.
[33]Sharma A, Bestavros A, Matta I. DPAM:A distributed prefetching protocol for scalable asynchronous multicast in P2P systems[C]//24th Annual Joint Conference of the IEEE Computer and Communications Societies.2005:1139-1150.
[34]Cheng B, Jin H, Liao X. RINDY:A Ring Based Overlay Network for Peer-to-Peer On-Demand Streaming[C]. In:Proc. of UIC 2006:1048-1058.
[35]Chang X. Network Simulations with OPNET [C]. In Proceedings of the 31st Conference on Winter Simulation:Simulation-a Bridge to the Future. New York:ACM Press, 1999,1:307-314.
[36]徐雷鸣,庞博,赵耀.NS与网络模拟[M].北京：人民邮电出版社,2003.
[37]VARGA A.OMNeT++ discrete event simulation system user manual [EB/OL]. (2009-6-26)[2009-8-10] http://www.omnetpp.org.
[38]朱晓姝.OMNET++仿真工具的研究与应用[J].大连工业大学学报,2009,29(1)：62-65.
[39]Jared W, Sugih J. Inet-3.0:Internet topology generator. Technical Report, CSE-TR-456-02[R]. Ann Arbor:University of Michigan,2002.
[40]Baumgart I, Heep B, Krause S. OverSim:A flexible overlay network simulation framework[C]//Proceedings of 10th IEEE Global Internet Symposium (GI'07) in conjunction with IEEE INFOCOM 2007. Anchorage, AK, USA:79-84.
[41]刘家瑞.NAPSTER案与文件共享技术的版权责任[J]. Science Technology and Law,2004, 4:69-77.
[42]See Pamela Salmuelson. DRM {and, or, versus} Law? [M]. Communication of the ACM April 2000,46:41-45.
[43]See Scott Hervey. Future of Online Music:Labels and Artists [M].15 Transnat'1 Law,2002.
1 李英壮,颜彦.基于BT的视频点播系统的服务器负载分析.中南大学学报(自然科学版),2010年,41卷(S1期)：332-335.主办单位：中南大学。(论文所属章节：第三章)
[2]Tomoya K, Shigeki Y. Application of P2P (peer-to-peer) technology to marketing[C] Cyberworlds, Proceedings 2003 International Conference,2003:372-379.
[3]K P, Gummadi S, Saroiu S D. Gribble. A Measurement Study of Peer-to-Peer File Sharing Systems. Proceedings of Multimedia Computing and Networking, January 2002(MMCN' 02), San Jose, CA, USA.
[4]Clip2. Gnutella measurement project [M/OL].2001,5. http://www.clip2.com/.
[5]Nathaniel L, Matei R, Adam W. Wierzbicki. Deconstructing the Kazaa Network,3rd IEEE Workshop on Internet Applications (WIAPP'03) [C], Santa Clara, CA, USA,2003, 6:112-120.
[6]Chu Y H, Sanjay G, Rao S S. A Case for End System Multicast [J]. IEEE Journel on Selected Areas in Communications.2002,20(8):p.1456.
[7]Banerjee S, Bhattacharjee B, kommareddy C. Scalable application layer multicast[C]. Proceedings of the ACM SIGCOMM 2002, Pittsburgh, PA.2002,8.
[8]Deshpande H, Bawa M. Streaming Live Media over a Peer-to-peer Network [R].2001, Technical report, Stanford University.
[9]Deshpande H, Bawa M, Garacia H M. Streaming live media over peers [R]. Tech. Rep, Stanford University,2002,3.
[10]Guo Y, Suh K, Kurose J, et al. P2cast:Peer-to-peer patching scheme for VoD service [A]. In Proc. of the 12th Int'1 Conf. on World Wide Web[C], New York:ACM Press, 2003:301-309.
[11]Eager D, Vernon M, Zahorjan J. Minimizing bandwidth requirements for on-demand data delivery [J]. IEEE Trans. On Knowledge and Data Engineering,2001,13(5): 742-757.
[12]Padmanabhan V, Wang J, Chou A, et al. Distributing streaming media content using cooperative networking[C]. Proceedings of ACM/IEEE NOSSDAV. Miami, FL, USA.2002.
[13]Castro M, Druschel P, Kermarrec A, et al. SplitStream:High-bandwidth multicast in a cooperative environment [J]//Proceedings of 19th ACM Sysposium on operating System Principles. Lake Bolton, New York.2003,3:292-303.
[14]Rowstron A, Druschel P. Pastry:Scalable, Decentralized Object Location and Routing for Large-Scale Peer-to-Peer System [J]//Computer Science,2001(2218): 329-350.
[15]Zhang X, Liu J, Li B, et al. CoolStreaming/DONet:A data-driven overlay network for efficient live media streaming. [C]//Proceedings of IEEE INFOCOM,2005,3: 2102-2111.
[16]杨天路,刘宇宏,张文,等.P2P网络技术原理与系统开发案例[M].北京：人民邮电出版社,2007.
[17]Shi S, Turner J. Multicast routing and bandwidth dimensioning in overlay networks [J]. IEEE Journal on Selected Areas in Communications,2002,20(8):1444-1455.
[18]Do T, Hua T, Kien A. P2VoD:providing fault tolerant video-on-demand streaming in peer-to-peer environment[C]//Communications,2004 IEEE International Conference,2004,3:1467-1472.
[19]Gang W, Tzi-cker C. How efficient is BitTorrent? [C]//Multimedia Computing and Networking, San Jose, CA, USA,2006.
[20]Aguilar A, Ege R K, Makki K, et al. Enabling Peer Cooperation in Private Local Area Networks using BitTorrent[C]//Computers and Communications.2007,615-622.
[21]Mario B, Alfio L. An Analytical Model of a BitTorrent Peer [C]//Distributed and Network-Based Processing,2007:482-489.
[22]Comparison between the different BitTorrent tracker modes [EB/OL]. http://www.phpsoft.org/trackermodes.php.
[23]Thommes R, Coates M. BitTorrent Fairness:Analysis and Improvements[R]. Department of Electrical and Computer Engineering,2004,5.
[24]蒋雪岭.P2P流媒体系统多点下载技术的实现研究[J].沈阳大学学报,2006,18(4)：60-63.
[25]Bram Cohen. Incentives Build Robustness in BitTorrent. Workshop on economics of P2P system.2003,6.
[26]张杰,张尧,黄道颖,等.P2P网络中一种改进的片段选择控制策略[J].微计算机信息,2007,23(21)：140-142.
[27]Ying L, Yang L. Experimental Views on Neighbor Selection in BitTorrent[C]. Network and Parallel Computing Workshops NPC Workshops.2007:813-818.
[28]刘永卫,王东.BitTorrent系统中一种自适应阻塞算法[J].计算机工程与用,2007,43(34)：122-124.
[29]Dana C, Li D, Harrison D, et al. BASS:BitTorrent assisted streaming system for video-on-demand[C]//Multimedia Signal Processing,2005 IEEE 7th Workshop on Shanghai,2005:1-4.
[30]Vlavianos A, Iliofotou M, Faloutsos M. BiToS:Enhancing BitTorrent for Supporting Streaming Applications [C]//Computer Communications Processing, INFOCOM 2006. 25th IEEE International conference, Barcelona,2006:1-6.
[31]Qiu D, Srikant R. Modeling and Performance Analysis of BitTorrent-like peer-to-Peer Networks. Proc[C]. ACM SIGCOMM, Karlsruhe, Germany,2004,8: 367-378.
[32]郑伟平,齐德昱,徐克付.BitTorrent类型P2P流媒体系统研究[J].计算机科学,2010,37(2)：123-125.
[33]Sharma A, Bestavros A, Matta I. DPAM:A distributed prefetching protocol for scalable asynchronous multicast in P2P systems[C]//24th Annual Joint Conference of the IEEE Computer and Communications Societies.2005:1139-1150.
[34]Cheng B, Jin H, Liao X. RINDY:A Ring Based Overlay Network for Peer-to-Peer On-Demand Streaming[C]. In:Proc. of UIC 2006:1048-1058.
[35]Chang X. Network Simulations with OPNET [C]. In Proceedings of the 31st Conference on Winter Simulation:Simulation-a Bridge to the Future. New York:ACM Press, 1999,1:307-314.
[36]徐雷鸣,庞博,赵耀.NS与网络模拟[M].北京：人民邮电出版社,2003.
[37]VARGA A.OMNeT++ discrete event simulation system user manual [EB/OL]. (2009-6-26)[2009-8-10] http://www.omnetpp.org.
[38]朱晓姝.OMNET++仿真工具的研究与应用[J].大连工业大学学报,2009,29(1)：62-65.
[39]Jared W, Sugih J. Inet-3.0:Internet topology generator. Technical Report, CSE-TR-456-02[R]. Ann Arbor:University of Michigan,2002.
[40]Baumgart I, Heep B, Krause S. OverSim:A flexible overlay network simulation framework[C]//Proceedings of 10th IEEE Global Internet Symposium (GI'07) in conjunction with IEEE INFOCOM 2007. Anchorage, AK, USA:79-84.
[41]刘家瑞.NAPSTER案与文件共享技术的版权责任[J]. Science Technology and Law,2004, 4:69-77.
[42]See Pamela Salmuelson. DRM {and, or, versus} Law? [M]. Communication of the ACM April 2000,46:41-45.
[43]See Scott Hervey. Future of Online Music:Labels and Artists [M].15 Transnat'1 Law,2002.
1 李英壮,颜彦.基于BT的视频点播系统的服务器负载分析.中南大学学报(自然科学版),2010年,41卷(S1期)：332-335.主办单位：中南大学。(论文所属章节：第三章)