Towards a Metric for Reuse of Microservice Intrusion Detection Models
José Flora, Nuno Antunes · 2024
Application development based on microservices allows leveraging continuous delivery to efficiently reach the market. This architectural methodology facilitates automated testing and deployment of frequent service changes. Nevertheless, it is imperative to incorporate security into this paradigm and ensure its continuous adaptation to the evolving landscape. We explore host-based intrusion detection by capturing microservices’ behavior, which encounters a significant challenge when dealing with the dynamicity of microservices that causes profiles to become outdated. This paper presents a preliminary study to assess the feasibility of a profile reutilization metric. It leverages data from the service’s source code, its runtime data and algorithm patterns to understand whether models from prior versions can be repurposed to newer versions. We apply it to four service versions within a microservices application to validate if we are on the correct path towards a valid and useful metric.