A Tag-Based Scheme to Realize Real-Time File Search in Hierarchical Peer-to-Peer Systems

Abstract

Abstract:In this paper, we propose a new method to realize quick update of information concerned with shared contents in Peer-to-Peer (P2P) networks. The proposed method is a combination of a hierarchical P2P architecture and a tag-based file management scheme. The hierarchical architecture consists of three layers: the top layer consisting of a collection of central servers, the middle layer consisting of a set of sub-servers, and the bottom layer consisting of a number of user peers. Indexes of files held by each user peer are stored at the sub-servers in the middle layer, and the correlation between file indexes and sub-servers is maintained by central servers using tags. We implemented a prototype of the proposed method using Java, and evaluated the performance through simulations using PeerSim 1.0.4. The results of our experiments indicate that the proposed method is a good candidate for "real-time search engines" in P2P systems; e.g., it completes an upload of 10,000 file indexes to the relevant sub-servers in a few minutes and achieves query forwarding to relevant peers within 100 ms.

Load Balancing Schemes for a Hierarchical Peer-to-Peer File Search System

“…However, in order to handle hetero-geneity of peers, some of peers are selected to be index servers of files shared by ordinary user peers. These peers are called "super-peers" [26,25,50,53,61,95,98,103]. Hierarchical P2Ps such as Kazaa [45], which are based on the notion of super-peers, have attracted considerable attentions in recent years.…”

“…To overcome such a difficulty, the use of super-peers is currently being the most promising approach in optimizing allocation of system load to peers, i.e., allocate more system load to high capacity and stable super-peers by assigning task of index maintenance and retrieval to them. In this thesis, we focused on two kind of superpeer based hierarchical architectures of P2P systems, which are distinguished by the organization of super-peers [25,61]. In each of them, we discussed system load allocation, and proposed novel load balancing algorithms for alleviating load imbalance of super-peers, aiming to decrease average and variation of query response time during index retrieval process.…”

“…In this Chapter, we focus on a hierarchical P2P proposed by Qin et al [61]. This is a three-tier P2P architecture consisting of top, middle, and bottom layers, where indices of files held by the user peers (UPs, for short) in the bottom layer are stored at the super-peers (SPs, for short) in the middle layer, and the correlation between those two layers is maintained by the central server(s) in the top layer using the notion of tags (see Section 3.3.1 for the details).…”

“…As will be described later, in this system, an overloaded SP easily becomes a bottleneck of query processing, which would significantly degrade the overall performance of information retrieval. In other words, Qin's hierarchical P2P has a great potential for striking a good trade-off point between the search efficiency and the system cost if we could remove such bottleneck by balancing the load of SPs, although it was not explicitly addressed in [61].…”