Exploiting locality using geographic coordinates and semantic proximity in Chord

Abstract

Sharing of content on the internet has become more and more prevalent over the years. As the number of internet users grows exponentially, issues of scale related to content distribution and lookup have become more important. But, since both content and network node are assigned a unique identifier in the basic design of Distributed Hash Table (DHT), randomly hashed to a same id space, " locality-awareness" is not inherent due to the topology mismatching between the Peer to- Peer overlay network and the physical underlying network. In this paper, we propose to incorporate physical locality onto nodes as well as semantic approach onto content into a Chord system. Initially, we incorporate network positioning model for capturing physical location information of network nodes. Secondly, we organize the data resources based on the domain of interest to route requests to the most likely physical location of data. For this reason, we incorporate GNP (Global Network Positioning) into Chord since peers can easily maintain geometric coordinates that characterize their locations in the Internet. Besides, we apply the Topic maps approach as a semantic layer in organizing resources based on the domain of interest on top of the Chord Protocol. The measured results show that Semantic-Chord-GNP efficiently locates the nearest available node by reducing the number of visited nodes as the overlay hop number. In addition, both the query latency and the the average success rate for content location when using Semantic-Chord-GNP are less than the basic algorithm of Chord.© 2010 IEEE.