Temporal Acceleration for Cloud-CDN-Fog-Edge Hierarchy by Leveraging Proximal Object Replicas

Abstract

A classical requirement to reduce network latency drives paradigm shifts from cloud computing together with content delivery network to fog computing, and edge computing. These paradigms practically co-exist in a multi-tier structure. This paper presents novel approaches for accelerating the cloud service response times of a cloud-CDN-fog-edge layered structure by means of hierarchical cloud caches and fine-grained replication. We also propose a unified cost-performance model that enables both temporal and monetary cost evaluation. We leverage the model to quantify the temporal effectiveness of the approaches in comparison with a classical LRU one. Results show that, when deploying in conjunction with the fine-grained replication, cloud caches significantly improve the overall response times up to 23.85%. A cloud layer is found to contribute most to the response times. Regardless of selected approaches, another finding is that the response times are the quadratical functions of the number of edge devices.