Monitoring and Analysis of Workflow Applications: A Domain-specific Language Approach

The increased focus of companies on improving their operational efficiency has raised a demand for advanced systems to support the continuous monitoring and analysis of business activities. These business activities are typically automated into workflow applications which are modeled, implemented and executed by workflow management systems (WFMS). The monitoring and analysis solutions aim at giving organizations the opportunity to focus on process improvement by detecting problematic properties of workflow applications. The focus of this research is to raise the level of abstraction for workflow developers for monitoring and analyzing workflow applications during their execution. Contemporary solutions require a big effort from workflow developers to materialize the needs of workflow analysts. First of all, these solutions fall short because they do not treat monitoring and analysis as a first class entity in the workflow implementation. As a result, workflow developers need to build ad hoc infrastructures to instrument the workflow implementation with monitoring and analysis concerns. This results in crosscutting and entangled code that affects the maintainability of the workflow application and the monitoring and analysis implementation. Another consequence is that the implementation of monitoring and analysis concerns is not reusable across different workflow platforms. Secondly, workflow variables and measurement variables are difficult to localize, use and share between workflow developers. As a result, there is limited support to base the monitoring and analysis on the data of the workflow application domain (e.g., Banking). Thus, the expressiveness of monitoring and analysis concerns is limited to generic properties defined in terms of the operational state of the workflow engine (e.g., the time a workflow is running). This dissertation presents two main contributions to solve these problems. Our first contribution is the creation a domain-specific language named MonitA to specify monitoring and analysis concerns in a uniform and workflow technol-