Abstract

Cloud computing has emerged as a very important commercial infrastructure that promises to reduce the need for maintaining costly computing facilities by organizations and institutes. Through the use of virtualization and time sharing of resources, clouds serve with a single set of physical resources as a large user base with altogether different needs. Thus, the clouds have the promise to provide to their owners the benefits of an economy of calibration and, at the same time, become a substitute for scientists to clusters, grids, and parallel production conditions. However, the present commercial clouds have been built to support web and small database workloads, which are very different from common scientific computing workloads. Furthermore, the use of virtualization and resource time sharing may introduce significant performance penalties for the demanding scientific computing workloads. In this paper, we analyze the performance of cloud computing services for scientific computing workloads. This paper evaluate the presence in real scientific computing workloads of Many-Task Computing users, that is, of users who employ loosely coupled applications comprising many tasks to achieve their scientific goals. Our effective method demonstrates to yield comparative and even better results than the more complex state-of-the-art techniques but has the advantage to be appropriate for real-time applications.

Authors and Affiliations