M-Ring : 在超級節點網路上能提供負載平衡的分散且自我組織的通訊連結方法

M-Ring : A Distributed, Self-Organize, Load Balancing Communication Method on Super Peer Network

林宗翰(Tsung-Han Lin)

指導教授 ： 鍾葉青

國立清華大學/電機資訊學院/資訊系統與應用研究所/碩士(2007年)

https://doi.org/10.6843/NTHU.2007.00397

摘要

許多的同儕式檔案分享系統的設計上開始將節點的地域性與異質性考慮進來。二階層式結構即是將不同的節點分群以及選擇少數幾個能力強的節點作為超級節點來達成與利用上述兩種性質。為了將鄰近分群內的節點互相通訊，鄰近群內的各個超級節點必須互相知道彼此資訊，或是各群內要有少數幾個當作閘道點來連結不同的分群。使用閘道點來做分群間的通訊可能會造成單點失敗(single-point-of-failure)或是使用讓各個超級節點知道彼此的方法則會造成極大的負擔。在此篇論文內，我們介紹M-Ring: 在超級節點網路上能提供負載平衡的分散且自我組織的通訊連結方法。M-Ring將所有的超級節點在同一個命名空間內給予一個獨特且唯一的識別碼。在同一命名空間內我們觀察到了一個"操作區域重疊特性"，並藉由此種特性來提升在超級節點上鄰近分群間的搜尋效率。我們的方法藉由各個超級節點僅儲存部分資訊可以節省各個超級節點整體的儲存空間與網路頻寬，而模擬結果也表現出我們的方法可以在超級節點之間的連結上有效節省維護負擔。

關鍵字

超級節點網路 ； 群組通訊 ； 操作區域重疊特性

並列摘要

Many peer-to-peer file sharing systems have been proposed to take locality and heterogeneous in to account. Two-layer architecture is one of the most used systems with these abilities by classified peers into groups and using some powerful peers as super peers. In order to communicate with other super peer set in its neighboring groups, each super peer has to know all other super peers in its neighboring groups or through some gateway-like super peer to connect to other groups. It may cause a single-point-of-failure if use gateway peer and may also cost a huge overhead if use the full-connect communicate method. In this paper, we propose M-Ring, a distributed, self-organized, and load balancing communicate method between each super peer set. M-Ring let each super peer constructs its own link table to connect to other super peer sets in its neighboring groups. M-Ring made each super peer in different groups has a uniquely identity id in the same identity space. We observed a “handle scope overlap” feature when all super peers in different groups are in the same identity space and use this feature to enhance the efficiency of query hops of super peer set in each neighboring group. Our method can reduce the total storing space and bandwidth in each super peer by stores differ part of all metadata in each super peers. The simulation results demonstrate that our approach provides an efficiently maintaining overhead on each connection to other super peers in neighboring groups.

並列關鍵字

super peer network ； group communication ； handle scope overlap

參考文獻

[1] Michael J. Freedman, and David Mazi`eres, “Sloppy Hashing and Self-Organizing Clusters,” In 2st International Peer To Peer Systems Workshop (IPTPS 2003), Berkeley, CA, USA, February 2003.

[13] Ion Stoica, Robert Morris, David Liben-Nowell, David R. Karger, M. Frans Kaashoek, Frank Dabek, and Hari Balakrishnan, "Chord: A Scalable Peer-to-Peer Lookup Protocol for Internet Applications," IEEE/ACM Transactions on Networking, VOL. 11, NO. 1, FEBRUARY 2003

[5] Gisik Kwon, Kyung D. Ryu, “An Efficient Peer-to-Peer File Sharing Exploiting Hierarchy and Asymmetry,” Proceedings of the 2003 Symposium on Applications and the Internet(SAINT’03), pp. 226–233, 2003

[6] Jian Liang, Rakesh Kumar , Keith W. Ross, “The KaZaA Overlay: A Measurement Study,” Computer Networks Journal (Elsevier), 2005.

[7] Virginia Lo, Dayi Zhou, Yuhong Liu, Chris GauthierDickey, and Jun Li, “Scalable Supernode Selection in Peer-to-Peer Overlay Networks”, Proceedings of the 2005 Second International Workshop on Hot Topics in Peer-to- Peer Systems (HOT-P2P’05), OR, USA, July, 2005