A Model Traceability Framework for Network Service Management

Abstract

Automating the enactment of processes using model-driven methods and tools paves the way for streamlining or optimizing these processes. Establishing traceability in automated processes is instrumental in carrying out analysis of the process and the involved artifacts.In this thesis, we propose a traceability information generation, visualization and analysis approach integrated with process modelling and enactment. A process model (PM) defined as an Activity Diagram has associated model transformations implementing the various activities and actions in the process. Enactment of the PM is carried out with the use of model transformation chaining in cooperation with model management means, in particular, megamodelling. We have incorporated both traceability in the small (at the model transformation level) and traceability in the large (at the PM level) in our approach. The traceability information is retained in the megamodel and forms the basis for traceability analysis of the enacted process. We have built a change impact analysis which allows the impact of a change in a model involved in the process to be assessed with the help of the derived megamodel.We further extended our approach with the notion of intents. We propose the usage of intents at both the PM and model-transformation levels as part of our traceability information. We define intents as information representing the objective of the PM actions/activities and their implementations. Furthermore, we have incorporated traceability visualization support to visualize trace links relating models at different levels through the captured intents. The intent-enriched traceability information and the enhanced visualization enable semantically richer traceability analysis.We applied our work to Network Service (NS) management in the context of the Network Functions Virtualization (NFV) paradigm.We believe automation of the orchestration and management of network services can progress rapidly with the help of model-driven engineering methods and tools. We applied our approach on a NS design process to analyze the impact of changing input models on output models as well as to show the benefits of intents not only in the context of this process, but also for the whole NS lifecycle management operations.Our work is concretized in a tool, MAPLE-T, built as an Eclipse plugin. It extends MAPLE, an integrated process modelling and enactment environment.