Authors:
Kun Huang
and
Dafang Zhang
Affiliation:
Hunan University, China
Keyword(s):
Computing infrastructure, Peer-to-Peer network, Distributed hash table, Broadcast, Load balancing.
Related
Ontology
Subjects/Areas/Topics:
Grid and Cluster-Based Computing
;
Mobile Software and Services
;
Peer-To-Peer Computing
;
Telecommunications
;
Wireless Information Networks and Systems
Abstract:
Scalable and efficient broadcast is essential to the large-scale computing infrastructures such as PlanetLab and Grids. By exploiting the greedy routing mechanisms of Distributed Hash Table (DHT), existing DHT-based broadcast algorithms suffer from the limitations of scalability and load balancing, incurring high construction and maintenance overhead for a distributed broadcast tree (DBT). This paper presents a token-based broadcast algorithm over DHT for the large-scale computing infrastructures, where each node selects the finger nodes as its children by a token value in a top-down approach. Theoretical analysis and experimental results show that the token-based broadcast algorithm can construct and maintain a balanced DBT with low overhead, where the branching factors of each node are at most two, and the tree height is in a Chord of nodes, without any extra storage space and explicit maintenance overhead.