A-DECS: Enhanced collaborative edge–edge data storage service for edge computing with adaptive prediction

Abstract

Due to the rapid development of IoT and 5G technologies, end-users of the Internet are generating a sea of data, which burdens the cloud cores and data centers. The urgent traffic demand causes a bad quality of experience in the traditional reactive networks. With the improvement of hardware, the edge of the Internet can share some responsibility for cloud servers. In small-scale scenarios, offload computing and storage tasks to edge nodes can help improve the user experience and utilize resources of the edge networks. In this paper, we propose a collaborative edge–edge data storage service with adaptive prediction called A-DECS based on our previous work DECS (Zhou and Chen, 2020) in local area scenarios. It manages nodes in the edge network and fully utilizes the limited resources in the cluster. A-DECS picks the most appropriate node to offload tasks. It can also proactively replicate popular data and generate forwarding rules in advance. A-DECS can reduce the read latency of cold-start and sudden peak flow situations. We compare it to state-of-the-art research. This experiment result proves that A-DECS is more suitable for edge clusters due to its lower latency and better resource utilization.