A GPU Accelerated High Performance Cloud Computing Infrastructure for Grid Computing Based Virtual Environmental Laboratory
Francisco Giunta, Raffaele Montella, Giuliano Laccetti, Florin Isailă, Francisco Javier Garca Blas · InTech eBooks · 2011
Advances in Grid Computing 122 set up an experimental infrastructure for the evaluation of scientific ideas.Our Virtual Laboratory has the ambitious goal to spread elastic and virtually unlimited computing power to scientists belonging to different e-science communities in a transparent and straightforward fashion.The rest of the chapter is organized as follows: section 2 describes the design and implementation of the application orchestration component.Section 3 is dedicated to the GPU virtualization technique, to the description of key design choices and of some benchmark results, which demonstrate the light impact of our software stack on overall performance; in that section, we also discuss about some high performance cloud computing scenarios depicted by recent gVirtuS enhancements.The need of a high performance file system distributed among virtual machines instances is the main motivation of the section 4; we discuss the design and the implementation of a virtual distributed file system mainly targeted to achieve high performance in dealing with HDF and NetCDF data archives.Our attention on those hierarchical self describing data formats is due to their extensive use in computational environmental science.Hence, section 5 sums up the virtual laboratory software ecosystem and describes the testbed application we have used in order to trim up the implemented features.Finally, the conclusions and future direction in this field are reported in section 6, where the principal guidelines for improving the usability and the availability of the described software components are drown.