A cache-node selection mechanism for data replication and service composition within cloud-based systems

Abstract

Data replication services provide clients with faster application access time. Since cloud applications generate tremendous amounts of stored information, data distribution is a must. Such distribution requires data to be replicated on third-party cloud storage sites for faster access. Current replica services exhibit an increase in data acquisition by clients and thus requires a partial file caching strategy. This paper proposes a mechanism for selecting mobile nodes to partially cache highly requested files. The proposed scheme relies on candidate mobile cache nodes that can efficiently provide file access services to other nearby nodes. The candidate cache nodes are chosen according to their devices' hardware and software resources, flow time, residual power, mobility characteristics and transmission capabilities. Additionally, services are composed on-demand using a fuzzy-induced service-specific overlay composition technique. Simulation results demonstrate the significant gains achieved by the proposed scheme in terms of access time, enhanced service availability, overlay composition delay reduction and high file hit ratios.