Scalable and dynamic replica consistency maintenance for edge-cloud system

Abstract

With the increasing number of edge devices in the Internet of things (IoT), data storage in edge cloud environment better meet the real-time requirements of IoT. In order to improve the data reliability, reduce the overhead and the response latency, and make more reasonable use of storage resources in edge cloud environment, we conduct an in-depth study on the replica creation strategy and consistency maintenance strategy. In order to reduce the waste of storage resources, a dynamic replica creation strategy based on block popularity and node load (DRC-BN) is proposed. DRC-BN strategy considers the storage cost of edge cloud and the communication cost between central cloud and edge cloud when optimizing the number of replicas, and improves the data reliability. In order to solve the problem of data inconsistency caused by multi-user concurrency, a replica consistency maintenance strategy based on Fast Paxos algorithm (RC-FP) is proposed. RC-FP strategy considers the comprehensive performance of nodes and selects the node with better performance as the leader according to the comprehensive performance of nodes, and reduces the time of consistency maintenance. Experimental results show that the proposed strategies perform well in terms of the consistency maintenance time and the node throughput. At the same time, the proposed strategies also reduce the overall overhead.