Cooperative Load Balancing of Neighboring Edge Clouds with LISP
Toru Furusawa, Akihiro Nakao · 2024
Cooperative load balancing effectively prevents ser-vice quality deterioration, particularly under elevated loads in edge offloading services of resource-limited edge clouds. It does this by transiently dispersing the load across multiple geographically proximate edge clouds. Our study introduces a method that enables quick load redistribution among adjacent edge clouds, facilitating rapid load distribution through the dynamic redirection of requests from terminals based on the load status of the respective edge cloud. This is achieved by updating the route information within the relay network. To validate our approach, we developed a prototype of an edge cloud infrastructure using Kubernetes and a controller responsible for routing control between edge clouds utilizing Locator/Identity Separation Protocol (LISP). Through experimentation, we demonstrate an 80 % reduction in median response time when distributing the load across neighboring edge clouds.