Designing a Distributed Design Exploration Framework in the Inter-cloud Environment

We propose an extreme-scale distributed design exploration framework in the inter-cloud infrastructure consisting of supercomputers and cloud computing systems. In design explorations, we need to search optimal configurations of input parameters that maximize or minimize objective function(s). In our framework, simulations are performed in parallel employing multiple supercomputers of the Japanese nation-wide high-performance computing infrastructure, and virtual or real machines in academic cloud systems are employed for parameter surveys and optimizations with machine learning. To enable nation-wide scale collaborations of supercomputers and cloud systems, we employ scalable distributed databases and objective storages to store input design parameters and resulting output information generated by simulations.