Deployments of micro-services to Kubernetes clusters can consist of a variety of implementations and their interaction can result in very complex system behaviour. Chaos Engineering is a testing strategy that exposes these deployments to a variety of stress tests in order to reveal potential bottlenecks and system failures.
Current approaches to Chaos Engineering provide functionality for scripting of sets of individual tests; however, the area lacks coherent strategies that ensure that the space created by the available variables is explored comprehensively and that deployments are exposed to an adequate number of tests while avoiding to have to execute excessive numbers of tests.
This project will investigate the development of an approach that allows engineers to configure the level to which the variable space is explored, potential emphasis of tests and the level of distribution of the individual tests.