High availability is a key requirement for many modern cloud computing systems, in particular, in conjunction with the Internet of Things. Therefore, the University of Stuttgart and Robert Bosch GmbH have created this joint research project that aims at novel modeling and prediction methodologies for predicting the availability of cloud applications during the development phase. The results of this project should help developers in developing highly available cloud applications in a cost efficient manner, while maintain service level requirements at any stage of the development phase. In order to cope with the high complexity of sophisticated cloud systems, we use probabilistic graph models as our mathematical framework to predict the availability of a cloud applications. Moreover, we introduced Clams (Cloud application modeling solution), a novel open source cloud-modeling language, to meet the modeling demands of the developer to describe their architecture.
Clams is a cloud modeling language to describe cloud applications as component-based system through well-known Cloud Computing Patterns in a combination with sequence diagrams. Clams imposes a new architectural design paradigm called “Design by Intention” (DbI), where the developer can state his architectural design intentions rather than explicitly use concrete service offerings. DbI allows for architectural uncertainty while maintaining the character of the architecture. In a plug-and-play fashion, recommendation algorithms can then be used to propose concrete services that meet the demands of the developer. Clams is open source and available as OpenClams at GitHub.