We develop different cost models for fundamental elements, covering virtual machines, storage, I/O, and networks and for fundamental execution models. We breakdown application execution at runtime into different parts associated with cost models and determine costs based on monitoring data, application structures and resoure usage. We take into account runtime compute resource capabilities and performance, application and system performance, and expected cost and quality. In our approach, application execution lifecycle can be slicedinto different steps and different decision making algorithms will be applied for different parts that deal with elasticity tradeoffs w.r.t. cost, quality, and resource. We will build multi-dimensional elastic properties for data resources based on storage elasticity, data concerns and the concepts of data marketplaces. We develop algorithms to allow runtime decision making in data resource selection and provisioning for complex applications. We will utilize runtime quality, cost and resource information as well as predictive models for making automatic application refinements and execution based on different elastic trade-off models.
SmartSociety: "Hybrid and Diversity-Aware Collective Adaptive Systems: When People Meet Machines to Build a Smarter Society "
In the FP7 FET SmartSociety project, started in 2013, we will research and develop programming paradigms for hybrid and diversity-aware collective adaptive systems. Furthermore, we will also investigate incentive and rewarding mechanims for these systems. Check the web site of SmartSociety for further information.
U-Test: "Testing Cyber-Physical Systems under Uncertainty: Systematic, Estensible and Configurable Model-Based and Search-Based Testing Methodologies"
The H2020 U-Test project has started in 2015 and it aims at: i) Providing a comprehensive and extensible taxonomy of uncertainties, classifying uncertainties, their properties, and their relationship; ii) Creating an Uncertainty Modeling Framework (UMF) to support modeling uncertainties at various levels (relying on exiting modeling/testing standards; iii) Defining an intelligent way to evolve uncertainty models developed using UMF towards realistic unknown uncertainty models using search algorithms; and generating cost-effective test cases from uncertainty and evolved models. For furter information, please check the U-Test home page.