Date of Award
Master of Science
Dr. Jeanne Stynes
Dr. Reinhold Kröger
Knowing which workflows are executed within Service Oriented Architectures (SOA) is essential for successful IT management. In many cases, SOAs grew out of previous existing IT architectures; existing components are used as single services and therefore as parts of newly created workflows. Since such workflows consist of newly developed and legacy services, traditional workflow management systems often cannot be applied. This thesis presents a method for gathering information about the executed workflows within such heterogeneous environments. An implementation of a framework is presented. This framework allows the training of machine-learning algorithms with workflow models and the mapping of low-level monitoring information produced by Log4J back to appropriate workflows. The framework allows us to compare the results using machine-learning algorithms simulating different scenarios. The method presented and its prototypical implementation facilitate successful IT management within heterogeneous SOAs.
Stein, Thorsten, "A Machine-learning Approach for Workflow Identification from Low-level Monitoring information" (2011). Theses [online].
Available at: https://sword.cit.ie/allthe/104