Data Replication on the Cloud/Edge

Abstract

This work presents a database replication system capitalising on hybrid cloud/edge infra-structures, which may be used in novel software architectures like microservices' applications. The system aims to reduce the latency perceived by clients performing read and update operations on a database, by locating replicas on edge nodes nearer end users. The replicas' convergence algorithm is based on eventual consistency and presents a novel solution that combines Operations Transformation and CRDTs techniques. The system follows the MongoDB data model and is validated on geographically disperse Amazon AWS data centers simulating edge nodes.