Misuse Detection and Response for Orchestrated Microservices Based Software

In the evolving landscape of cloud computing, containerized microservices have emerged as a dominant architecture, presenting unique security challenges. This paper introduces a novel security framework, harnessing the power of machine learning, to enhance the detection and response capabilities against misuse in Kubernetes-based microservices environments. Central to our approach is the Dynamic Topology Adjustment (DTA) operator, seamlessly integrated with Kube-OVN’s advanced networking features, enabling proactive and dynamic adaptation of the network topology in response to real-time security threats. We implement an AI-driven misuse detection model based on the SGDOneClassSVM algorithm, tailored to analyze network flows within these complex systems. Our framework not only addresses immediate security concerns but also sets a foundation for adaptive, intelligent security management in cloud-based microservices. Experimental results, derived from a specially curated dataset targeting container-specific vulnerabilities, demonstrate the efficacy of our approach in detecting a range of security threats with high accuracy, showcasing its potential as a robust solution for container security in cloud environments.