Automated Experiments for Deriving Performance-relevant Properties of Software Execution Environments

In software engineering, considering quality attributes such as software performance plays a crucial role. Approaches such as Software Performance Engineering (SPE) aim at systematically reflecting software performance during the whole software life-cycle. Using model-based performance analyses, predictions on the software performance, e.g. the expected software response times or the resource utilization, can already be conducted during design time. The software execution environment heavily influences the performance of a software and thus has to be reflected in performance analysis. With the increased complexity of the execution environment (e.g. due to the introduction of virtualized environments or new operating system schedulers), performance analysis approaches have to be extended in order to continue to yield precise prediction results. Existing performance prediction approaches usually abstract from most parts of the execution environment or do not deal with the problem of quantifying properties of the execution environment for performance prediction models. Hence, detecting execution environment properties and integrating such properties into performance analyses is a manual, error-prone task that requires expert knowledge on the execution environment. The scientific contribution of this thesis is a novel approach for detecting performance-relevant properties of the software execution environment. These properties are automatically detected using predefined experiments and integrated into performance prediction tools. To detect the properties, performance experiments are conducted on the target platform. The experiments issue predefined load patterns on the system and observe certain per-