Enterprise Application Discovery and Monitoring Management with ExplorViz
Alexander Krause · 2018
Developers’ knowledge and comprehension about program internals does not scale with the complexity and scope of today’s software. The vast number of lines of source code and external dependencies are two factors, which exacerbate this observation. Static analysis approaches, such as utility tools of integrated development environments, may reduce the number of overall problems. However, only dynamic analysis and monitoring tools are able to show flaws during program runtime. Such tools use different visualizations to depict analyzed data and potential software issues. Unfortunately, the underlying monitoring approaches often lack a plausible usability and require an unnecessary long time to setup. This might discourage developers from using these tools. Therefore, the latter miss their major purpose, i.e., improving software. In this thesis, we present an approach for an application discovery and monitoring management system, abbreviated to ADAMMS. Our developed design addresses critical features such as monitoring setup and application lifecycle or monitoring management. We implement the design in the form of an extension to the ExplorViz research tool, which explores software visualizations for dynamic runtime data. The resulting ExplorViz ADAMMS is able to discover running Java applications in Debian operating systems. The software facilitates the implementation of custom discovery mechanisms, based on a provided scaffold. Thus, our extensible system is capable of supporting different types of applications and operating systems. In a subsequent evaluation, we successfully show the feasibility of our implementation. We further examine the usability by means of a pilot study, which includes a task-oriented experiment and a structural interview. The results indicate a good usability, but further experiments with more probands are necessary to obtain comparable results. Both experiments reveal small adjustments that would possibly enhance the implementation. Finally, we give an outlook and suggest topics for future work.