P2P 影音串流服務之可擴充性遞送系統研製

近年來，隨著網路影音串流應用的服務和需求量日益增加，高畫質且穩定的影音串流傳輸品質逐漸成為眾多服務平台的共同目標與課題之一。因此使用點對點網路傳輸(P2P Network)技術以降低高畫質影音傳輸時所需的網路頻寬和服務伺服器負載量即成為了一個熱門的議題。然而為了讓在影音平台上之使用者皆能擁有良好的串流服務品質，P2P串流伺服器必需花費大量的硬體(CPU，RAM等)和網路頻寬資源以動態維護P2P網路拓樸(Topology)的完整性。由於單一台伺服器之硬體和網路環境僅能負荷服務一定數量的使用者，倘若觀看人數超出單台串流伺服器之最大負荷人數，串流提供者必須準備或租用多台伺服器以分散其大量使用者所產生的負載需求。雖然現今的雲端技術可以透過花費些許租金以取得其虛擬機器以減少硬體設備的成本需求，但倘若租借過多的伺服器而線上的使用者人數不如預期，則會導致服務成本和資源上的浪費。對於串流服務提供者而言，很難在服務開始之前完全掌握其欲接受服務的人數規模，如何即時且彈性地依照目前的使用人數規模來調整伺服器的使用數量就成了一個不可忽視的議題之一。在此篇文中，我們提出一個能在P2P網路環境中達到負載平衡的即時管理系統以根據不同的使用者人數和網路情況對伺服器的使用做適當地調整。首先，我們建立一個管理伺服器以便隨時監控和更新每一台串流伺服器於目前的資源使用數據，並且掌握目前可以提供服務之伺服器的最新情況。進一步地，我們於管理伺服器上設計一套新的串流伺服器負載評估方法以計算每一台串流伺服器的資源使用量數據來分別其伺服器的負載程度。除此之外，我們在管理伺服器上提出一套演算法以對每一台P2P伺服器的負載程度資料進行處理和分配使用者至P2P伺服器。為了驗證我們所設計出來的系統，在本篇論文中我們也在實際網路環境中設計了許多不同的測試實驗並透過分析實驗結果來驗證我們系統的可行性和完整性。

關鍵字

網路點對點傳輸；負載平衡

並列摘要

Video streaming applications have become increasingly widespread in today’s Internet. The high-definition (HD) video technology has also been applied for the speech and ball game live broadcasting. For reducing the heavy bandwidth requirement, Peer-to-Peer (P2P) transmission technology is widely applied to video streaming services. The most advantage of P2P transmission is the ability to conserve bandwidth and reduce server load. Nevertheless, it is difficult to correctly and efficiently maintain the P2P network topology when a large number of users join to the group. Moreover, the streaming server’s loading is heavy in order to keep the quality of services (QoS) with a large number of users. Although streaming providers can solve this problem easily by renting more hardware servers or virtual devices, the streaming broadcast cost is also increased by using more streaming servers. In this thesis, a novel central control system is proposed to efficiently manage the P2P streaming servers to provide scalable and stable streaming service for a large number of users. First, a managing device called “manager server” is proposed to monitor the condition of each streaming server. Next, a loading condition evaluation method is implemented on the manager server to analyze each P2P server’s resources utilization. Moreover, two load balancing operations are offered to dynamically partition the peers into P2P servers to keep the whole system and service as stable as possible. Various experiments are conducted to evaluate the performance and scalability of the proposed system. Experimental results indicate that the proposed system architecture and the load balancing mechanisms work well for 2Mbps video streaming source with different number of joining peers.

並列關鍵字

P2P Network ； Load-balancing

參考文獻

[5] Chuan Wu, Baochun Li and Shuqiao Zhao, “On Dynamic Server Provisioning in Multichannel P2P Live Streaming”, IEEE/ACM Transactions on Networking (Volume:19 , Issue: 5 ), Oct. 2011, pp. 1317 – 1330.

[8] Ginhung Wang and Kuochen Wang, “An Efficient Hybrid P2P MMOG Cloud Architecture for Dynamic Load Management”, 2012 International Conference on Information Networking (ICOIN), 1-3 Feb. 2012, pp. 199 – 204.

[9] Ming Xu and Jihong Guan, “Routing Based Load Balancing for Unstructured P2P Networks”, Future Generation Communication and Networking (FGCN 2007), 6-8 Dec. 2007, pp. 332 – 337.

[10] Yingwu Zhu and Yiming Hu, “Towards Efficient Load Balancing in Structured P2P Systems”, Parallel and Distributed Processing Symposium, 2004. Proceedings. 18th International, 26-30 April 2004, pp. 1-10.

[11] Di Wu, Ye Tian and Kam-Wing Ng, “Achieving Resilient and Efficient Load Balancing in DHT-based P2P Systems”, Proceedings 2006 31st IEEE Conference on Local Computer Networks, 14-16 Nov. 2006, pp. 115 – 122.

國際替代計量

P2P 影音串流服務之可擴充性遞送系統研製

全文下載

主題瀏覽