The increasing complexity of cyber attacks requires regular training through Cyber Security Exercises (CSEs). This applies not only to operational environments but also to educational contexts, where structured scenarios support learning objectives. However, existing approaches to scenario modeling often show limits when it comes to flexibility, reusability, and technical integration into containerized environments. As part of this work, the Scenario Modeling Language (SML) is developed. It is a domain-specific modeling language based on a modular information flow approach. Methodologically, the work follows the Design Science Research (DSR) paradigm. A metamodel defines key elements such as injects, events, and gates which are represented graphically and in a machine-readable format (JSON). The approach is illustrated using a scenario from the CONTAIN research project and compared against functional and non-functional requirements. The results indicate that the SML can reduce modeling complexity and supports interoperability. The SML provides a structured way to describe and reuse scenarios and can support both training and teaching settings. Further work is needed to evaluate its use in more complex scenarios and environments.