Transformation of UML Deployment Models into executable Cloud Orchestration Templates

Johannes Erbel


In the recent years cloud computing has been increasingly used to dynamically assign com- puting resources on demand. Due to different tool and Application Programming Interface assistance, it is difficult to transfer cloud applications from one to another provider. Mean- while, there are several cloud orchestrators and standards solving this problem, requiring specific models. Because these are new standards, there are no proper tools supporting them. However, with transformations, a common method in Model Driven Engineering, a broadly used and well-supported modeling language can be utilised to generate a domain specific model, supporting the cloud standard. This thesis shows that model transfor- mation can transform models containing whole infrastructures into executable artifacts. This thesis shows that it can be used to transform models containing whole infrastruc- tures into textual artifacts. Therefore, the compatibility between the Unified Modeling Language (UML), a common modeling standard, and Cloudify, a Topology and Orches- tration Specification for Cloud Applications (TOSCA) based cloud orchestrator, is exam- ined. Each possible deployment model element is mapped to an element in the TOSCA model. Based on that mapping an automatic model transformation process from an UML deployment model to a TOSCA based model is created. Although not all information can be modeled in a deployment model, a large proportion can be transformed with an accu- rate modeling. Thereafter, the resulting TOSCA model is transformed into an executable Cloudify blueprint, enabling an automatic deployment of the modeled infrastructure.
Document Type: 
Bachelor's Theses
Gottingen, Germany
Institute of Computer Science, University of Göttingen
2024 © Software Engineering For Distributed Systems Group

Main menu 2