Cycloid: A constant-degree and lookup-efficient P2P overlay network

被引:68
作者
Shen, HY
Xu, CZ [1 ]
Chen, GH
机构
[1] Wayne State Univ, Dept Elect & Comp Engn, Detroit, MI 48202 USA
[2] Nanjing Univ, State Key Lab Novel Software Technol, Nanjing 210008, Peoples R China
基金
美国国家科学基金会; 美国国家航空航天局;
关键词
Cycloid; peer-to-peer; Viceroy; Koorde; distributed hash table; constant-degree DHT;
D O I
10.1016/j.peva.2005.01.004
中图分类号
TP3 [计算技术、计算机技术];
学科分类号
0812 ;
摘要
There are many structured P2P systems that use DHT technologies to map data items onto the nodes in various ways for scalable routing and location. Most of the systems require 0(log it) hops per lookup request with 0(log it) neighbors per node, where n is the network size. In this paper, we present a constant-degree P2P architecture, namely Cycloid, which emulates a cube-connected cycles (CCC) graph in the routing of lookup requests. It achieves a time complexity of O(d) per lookup request by using O(l) neighbors per node, where n = d x 2(d). We compare Cycloid with other two constant-degree systems, Viceroy and Koorde in various architectural aspects via simulation. Simulation results show that Cycloid has more advantages for large scale and dynamic systems that have frequent node arrivals and departures. In particular, Cycloid delivers a higher location efficiency in the average case and exhibits a more balanced distribution of keys and query loads between the node.,,. (D (C) 2005 Elsevier B.V. All Rights Reserved.
引用
收藏
页码:195 / 216
页数:22
相关论文
共 20 条
[1]   GROUP ACTION GRAPHS AND PARALLEL ARCHITECTURES [J].
ANNEXSTEIN, F ;
BAUMSLAG, M ;
ROSENBERG, AL .
SIAM JOURNAL ON COMPUTING, 1990, 19 (03) :544-569
[2]  
[Anonymous], 2001, UCBCSD011141
[3]  
[Anonymous], DISTRIBUTED DECENTRA
[4]  
CHEN G, 2003, LECT NOTES COMPUTER, V2975
[5]  
CLARKE I, 2000, P ICSI WORKSH DES IS
[6]  
FELDMANN R, 1991, PARALLEL PROCESS LET, V2, P13
[7]  
KAASHOEK MF, 2003, P 2 INT WORKSH P2P S
[8]  
LV Q, 2002, P ACM SIGGRAPH AUG
[9]  
LYNCH N, 2002, P INT PEER PEER S
[10]  
MALKHI D, 2002, P PRINC DISTR COMP P