Resource Utilization Comparison of KubeEdge, K3s, and Nomad for Edge Computing
Muhammad Bintang Bahy, Nur Rahmat Dwi Riyanto, Muhammad Zain Fawwaz Nuruddin Siswantoro, Bagus Jati Santoso · 2023
Containers have become increasingly popular as a lightweight form of virtualization technology in the past few years. This surge in popularity aligns closely with the growing use of microservice architectures, primarily due to the scalability, ephemeral nature, and isolation provided by containers. In more recent times, advancements in edge devices have enabled them to support the execution of containerized microservices. These devices maintain a desirable balance between their size and power capabilities, to be implemented in a variety of locations. Research into various container placement strategies within edge networks has been prompted, resulting in the emergence of concepts like osmotic computing. Although these strategies for placing containers are efficient in terms of weight allocation, container orchestrators at present typically demand high resources requirement running on edge devices which have limitations of capabilities in terms of managing processes.This article presents a performance comparison of several container orchestrations consisting of KubeEdge, K3s and Nomad. The comparative aspects include resource utilization like CPU, memory and storage use before and after the deployment process. The experiment’s evaluation indicates that orchestrators favour Nomad because of its exceptional ability to efficiently utilize CPU and memory resources. On the other hand, K3s stands out for its higher efficiency in utilizing storage resources compared to other orchestrators.