Investigating the Use of Graph Databases for Large Model Repositories

Kiel, Benjamin (2013) Investigating the Use of Graph Databases for Large Model Repositories (Master thesis), Kiel University, Kiel, Germany, 77 pp


Download (2583Kb) | Preview


In the context of model-driven reverse engineering and model-driven analysis of software systems, models can become very large. Up to a certain point, these models fit into memory, but often the available memory is not sufficient. Model repositories like the Connected Data Objects model repository (CDO) in the context of the Eclipse Modeling Framework (EMF) provide support for storing models and allow to work with multiple concurrent users on these models. In general, models are represented as object graphs. This leads to the assumption that graph databases could be applicable for storing models. This thesis investigates the usage of graph databases in the context of model-driven engineering. It provides a survey of available model repository technologies in the context of EMF. The survey is based on requirements for the MAMBA (Measurement Architecture for Model-Based Analysis) framework for model-based analysis. We introduce our implemented graph database backend for CDO. Furthermore, this thesis proposes a benchmark that can be used to examine the performance for CDO stores with respect to the required times for storing, loading, and querying models. Finally, we evaluate available CDO stores as well as our implementation with the help of this benchmark. We show that a graph database backend does not provide any advantage compared with a relational database backend in the context of CDO with respect to execution times.

Document Type: Thesis (Master thesis)
Thesis Advisors: Hasselbring, Wilhelm, Frey, Sören, Jung, Reiner, van Hoorn, Andre and Stahl, Thomas
Keywords: Model Repository, Graph Databases, CDO, Neo4j
Research affiliation: Kiel University > Software Engineering
Date Deposited: 14 Jun 2013 10:32
Last Modified: 22 Jul 2013 08:20

Actions (login required)

View Item View Item

Document Downloads

More statistics for this item...