Integrating Model Transformation Systems and Asynchronous Cluster Tools

Our primary research focuses on creating an efficient model transformation method that is based on graph transformations. We have found that there are cases when using single computers to compute the transformation results exhibits unacceptably slow transformations. Building clusters from computers is a well-known method to increase the computing capability in a cost-effective way. Thus, we focus on creating a distributed model transformation method based on cluster systems. In this case, the efficient communication between the computers in the clusters is essential. Therefore, a cluster management system has been chosen, which supports asynchronous, message-based communication. Cluster systems and distributable tasks in these systems are usually based on the C++ language, while our model transformation engine uses C#, which means that interoperability is required between the two fields. This paper describes the integration issues in our solution including an architectural overview and detailed information about the interoperability method.