Geoffrey Fox

Clouds have substantial growth in commercial applications but use in science is so far limited. We look at cloud systems (Azure) and platforms (MapReduce and its extensions) for data analysis applications - especially in genomics and proteomics. We discuss two difficulties - synchronization (communication) overhead and security for clouds that counter their cost effective elastic properties but suggest that whereas large scale simulations will continue to run on classic supercomputers, data analysis (mining) is suitable for clouds. We discuss platforms that are portable across cloud and HPC systems.