CAJ | 학술논문

Object location is a fundamental problem in distributed system such as grid computing and peer-to-peer environment. Chord is one of the typical scalable object location algorithms for peer-to-peer network. The simplicity, provable correctness and provable performance of Chord make it an attractive option for distributed lookup. However its lookup performance is not satisfying. Based on the decentralized object location mathematical model. This paper proposes ChordPlus, an improved Chord algorithm, by enlarging dimension size of space M (M = 2 in Chord) and utilizing data redundancy technology. ChordPlus is a scalable, fault-tolerant, completely decentralized and self-organizing object location and routing algorithm for overlay network. Results from theoretical analysis and simulation experiments show that increasing the dimension size of space or neighborhood set length can enhance the routing performance of ChordPlus. In a simulation network with 5 000 nodes, the average routing path length of ChordPlus is only as many as half that of Chord algorithm.