Environmental Modeling for Automated Cloud Application Testing

Platforms such as Windows Azure let applications conduct data-intensive cloud computing. Unit testing can help ensure high-quality development of such applications, but the results depend on test inputs and the cloud environment's state. Manually providing various test inputs and cloud states is laborious and time-consuming. However, automated test generation must simulate various cloud states to achieve effective testing. To address this challenge, a proposed approach models the cloud environment and applies dynamic symbolic execution to generate test inputs and cloud states. Applying this approach to open-source Azure cloud applications shows that it can achieve high structural coverage.