Proactive Fault Tolerant Service Management in Edge Computing
Supriya Dilip Tambe, A. Antony Franklin · 2023
Edge computing technology facilitates the deployment of real-time services near users by reducing latency and increasing efficient bandwidth utilization compared to cloud computing. The edge servers are usually heterogeneous and resource-constrained. To effectively utilize these edge resources, it is required to efficiently deploy different real-time latency-sensitive edge services so that maximum users will serve. In a pay-per-use edge environment, service providers can not deploy all services on all available edge servers, which is costly for service providers. Further, due to hardware/software issues or overloading, any edge server may fail after service deployment and hamper critical edge services’ system performance. In such a failure-prone environment, designing fault-tolerant service deployment strategies is essential to reduce the impact of the failures. In this paper, we propose a two-stage method, which includes the optimal approach for selecting a primary server to increase user coverage and a heuristic approach for impact-based service deployment within budget to choose a backup server. These stages also verify the overloading of services and user requests on the edge servers to reduce edge failures. Several experiments are conducted to analyze the proposed approach on real-world data. Compared to various baseline approaches, our solution efficiently increases user coverage (15%) and reduces failure impact by having a backup server (25%) for most users to increase availability within the deployment budget.