Popularity-based data placement in erasure-coded Edge Storage Systems

Abstract The explosive growth of smart devices and mobile users has led to an unprecedented demand for efficient data storage and processing. Edge computing has emerged as a promising solution to address these challenges by bringing computation and storage resources closer to data sources. Edge Storage Systems (ESSs) are a specific type of edge computing system designed to store and serve data efficiently at the network edge. While replication-based ESSs are a simple and effective approach, erasure-coded ESSs offer superior resilience, performance, and storage efficiency. However, effective data placement in erasure-coded ESSs is a complex problem, as it involves balancing factors such as data popularity, network latency, and resource utilization. In this paper, we propose a novel data placement approach for erasure-coded ESSs that addresses these challenges. Our approach consists of (1) a data placement strategy that leverages a virtual-space-based placement technique and incorporates data popularity into the placement decision, and (2) a dynamic data retrieval strategy that retrieves data from nodes with less load (in terms of data requests to be served). We evaluate our approach through extensive simulations, considering various network topologies, user mobility patterns, and data access patterns. Our results demonstrate that our proposed approach suitably addresses the data placement challenges arising from using erasure-coded ESSs, while allowing to exploit their advantages with respect to replication-based ESSs in terms of both space efficiency and data availability.