Key-value storage system synchronization in peer-to-peer environments

Abstract

Data synchronization is the problem of bringing multiple versions of the same data on different remote devices to the most up to date version. This thesis looks into the particular problem of key-value storage systems synchronization between mobile devices in a peer-to-peer environment. In this research, we describe, implement and evaluate a new key-value storage system synchronization algorithm using a 2-phase approach, combining approximate synchronization in the first phase and exact synchronization in the second phase. The 2-phase architecture helps the algorithm achieve considerable boost in performance in all three major criteria of a data synchronization algorithm, namely synchronization time, processing time and communication cost, while still being suitable to operate in a peer-to-peer environment. The performance increase makes it feasible to employ database synchronization technique in a wider range of mobile applications, especially those operating on a slow peer-to-peer network.