Research On Data Replication Strategy In Peer-to-Peer Information Resource Center Networks

The information of information system in wide-area distributed environment presents some new feature such as huge magnitudes, distributed, dynamic, heterogeneous et al. The research background of this dissertation is Information Resource Center networks based, Information Sharing Platform(IRCISP). The IRCISP platform forms a global information sharing environment by adopting multiple Information Resource Center (IRC) architecture, and provides high effective information sharing service with multiple IRC peers'cooperation. IRC nodes form a unstructured peer-to-peer network.Data replication can improve data search performance and system load balancing and achieve improvement of data reliability and fault-tolerance. We take data replication strategy in peer-to-peer information resource center networks into research, includes: when, which, how many, where and replica management, and we also study IRC's data recovery problem based on data replication in peer-to-peer networks. The major contributions of this dissertation are:1. This dissertation proposes an access frequencies-based replication strategy for unstructured peer-to-peer networks. Data access records gathered at different time intervals has different weights in order to distinguish the importance between historical records, then we calculates data access frequency and puts data replicas into peers which in areas that access frequencies are higher. The result of experiment shows that compared with those normal replication strategies, our algorithm increases the success rate and has a better performance of average search size when reducing total replica amounts.2. This dissertation gives a data replication based information resource center data recovery method, a data recovery-oriented data replication management framework, a multi-areas replica management model. This dissertation also takes data recovery-oriented replication strategy into research, the simulation results show that data replication strategy is effective and feasible for data recovery.