Comparison of Lightweight Kubernetes Distributions for Edge Computing
Yakubov, Diyaz (2024)
2024
Master's Programme in Computing Sciences and Electrical Engineering
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
Hyväksymispäivämäärä
2024-12-11
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-2024120910859
https://urn.fi/URN:NBN:fi:tuni-2024120910859
Tiivistelmä
The escalating demand for real-time data processing in Internet of Things (IoT) devices has amplified the significance of edge computing, necessitating efficient, secure, and resilient deployment of applications on resource-constrained devices. Lightweight kubernetes distributions — k0s, k3s, KubeEdge, OpenYurt, and the standard Kubernetes (k8s) — extend container orchestration capabilities to edge environments, yet comprehensive analyses of their performance, resource efficiency, security, resiliency, and maintainability remain limited. This study presents an empirical comparative analysis of these distributions, focusing on key metrics such as CPU, memory, disk usage, throughput, latency, security compliance (using kube-bench), resiliency under network outages, and maintainability.
Utilizing a testbed of Intel NUCs and Raspberry Pi devices, we evaluated the distributions under varying workloads. The results reveal significant trade-offs between performance, resource consumption, security, resiliency, and maintainability. Specifically, k3s exhibited the lowest resource consumption and superior maintainability due to its simplicity but demonstrated lower security compliance compared to k8s, KubeEdge, and OpenYurt. k0s offered efficient performance under heavy workloads and ease of use but also lagged in security features. Kubernetes provided a balanced approach and excelled in data plane throughput and latency. KubeEdge and OpenYurt enhanced security features and reliability during network outages but increased complexity and resource consumption.
These findings offer valuable insights for developers and practitioners deploying kubernetes in edge environments. The study highlights the necessity of balancing performance, security, resiliency, and maintainability according to specific deployment requirements in resource-constrained settings. Recommendations are provided to assist in selecting appropriate distributions based on these critical factors.
Utilizing a testbed of Intel NUCs and Raspberry Pi devices, we evaluated the distributions under varying workloads. The results reveal significant trade-offs between performance, resource consumption, security, resiliency, and maintainability. Specifically, k3s exhibited the lowest resource consumption and superior maintainability due to its simplicity but demonstrated lower security compliance compared to k8s, KubeEdge, and OpenYurt. k0s offered efficient performance under heavy workloads and ease of use but also lagged in security features. Kubernetes provided a balanced approach and excelled in data plane throughput and latency. KubeEdge and OpenYurt enhanced security features and reliability during network outages but increased complexity and resource consumption.
These findings offer valuable insights for developers and practitioners deploying kubernetes in edge environments. The study highlights the necessity of balancing performance, security, resiliency, and maintainability according to specific deployment requirements in resource-constrained settings. Recommendations are provided to assist in selecting appropriate distributions based on these critical factors.