Abstract: Continuous Delivery and Continuous Deployment approaches have seen widespread adoption in the software industry. To harness such techniques effectively, close monitoring and detailed knowledge about the state of software in production is highly desirable. This thesis analyzes the JDownloader Immune System, a real-time monitoring and error detection mechanism developed for the open source download manager software JDownloader. It describes the mathematical model for error detection based on time series analysis and Holt-Winters-Forecasting. The thesis continues to provide insight on the architecture of the immune system and shows how it provides useful information to developers and users through state dashboards. Finally, it evaluates the effectiveness of the immune system compared to manual user reports. The thesis finds that error detection speed for severe issues is 16 times faster than through manual reports and critical bugs are more than four times more likely to be detected within the first 24 hours after their first appearance.
Keywords: Continuous software engineering, continuous deployment, continuous delivery, JDownloader
Reference: Michael Weber. The Benefits of continuous deployment evaluated using the JDownloader software. Master Thesis, Friedrich-Alexander-Universität Erlangen-Nürnberg: 2017.