Efficient Docker-Based Resource Management for CPU-Intensive Tasks in Edge-Cloud Computing

Edge-cloud computing integrating the advantages of edge and cloud to provide high-quality services for mobile users is one of the hotspots recently. However, the increasing number of applications brings various tasks with different resource requirements, potentially leading to poor computing performance in Edge-cloud systems with fixed resource configurations, especially for tasks that are heavily reliant on CPU resources. To enhance computing performance, we first characterize the correlation between computing delay and CPU resources for CPU-intensive tasks based on Dockers technology. Then, we investigate a Docker-based resource management framework to handle task allocation, resource collaboration, and configuration for the edge-cloud computing system, where edge servers can allocate tasks to the cloud and purchase collaborative resources from the cloud while adjusting the configured number of Dockers. In such a way, edge servers can maximize their utility while ensuring quality of service. Based on the intersections between edge servers and the cloud, we formulate a utility maximization problem for edge servers that involves the expected allocated tasks to the cloud, collaborative CPU resources, and the configured number of Dockers, which is NP-hard. Next, based on the correlation analysis between the utility of edge servers and the optimized variables, we propose an effective task allocation and resource management scheme by alternating iterations to obtain the optimal solutions. Finally, we conduct extensive simulation results to show the performance of the proposed scheme in maximizing the utility of edge servers compared to other benchmarks.