Adaptive Controller to Identify Misconfigurations and Optimize the Performance of Kubernetes Clusters and IoT Edge Devices

Abstract

Kubernetes default configurations do not always provide optimal security and performance for all clusters and IoT edge devices deployed, affecting the scalability of a given workload and making them vulnerable to security breaches and information leakage if misconfigured. We present an adaptive controller to identify the type of misconfiguration and its consequence threat to optimize the system behavior. Our work differs from existing approaches as it is fully automated and can diagnose various errors on the fly. The controller is evaluated in terms of quality and accuracy of identification. The results show that the controller can identify around 90% of the total number of configuration values with a reasonable average identification overhead.