Gird or cloud? Survey on scientific computing infrastructure

Abstract

Recently, scientific computing not only brings benefits for science but also brings new challenges for scientists. Therefore, scientists have to face headache “4 how” problems, which are how to know, organize, use and guarantee resources, and as the solution, scientific computing infrastructure should coordinate and organize kinds of capabilities effectively, so as to provide on-demand and safe computing environment for scientists. So, this paper concludes the related research about scientific computing infrastructure in recent years, and base on that systematically analyze the primary demands of infrastructure, and moreover, we propose Capability Meta Model-CMM as the foundation model for kinds of infrastructure; and then, based on CMM we compare kinds of different infrastructure, including grid and cloud computing, from resources, knowledge, quality and security these four aspects, finally, we summarize the problems and challenges the current infrastructure is facing.