An application meta-model to support the execution and benchmarking of scientific applications in multi-cloud environments

Cloud computing has proven its importance to scientists around the globe on many occasions already. However, as it is still a rather new technology for many users, the cloud represents another layer of complexity in any workflow. As a lot of research confirms, especially the efficient provision and management of resources in the cloud is a very complex but also very rewarding task. Upon surveying the research in this area we observed many differences in applied methodologies and application cases which impede not only the comparison of these approaches but also the collective usage of the obtained results, e.g. for more accurate resource estimation algorithms that require less additional benchmarks. We propose a novel application and resource meta-model to model not only applications but also the underlying resource infrastructure for application benchmarks in a generic manner. We show how the meta-model is defined and how it can be used to model an application, using a simple web application as an example. We conclude with highlighting the potential benefits of applying this model in different scenarios but also its limits and how it could be expanded in the future.