Configurable performance analysis and evaluation framework for cloud systems

Abstract

Performance analysis and evaluation are considered as key infrastructures for distributed computing platforms, such as clusters and grids. Recently, cloud computing has become an emerging commercial infrastructure paradigm, which promises to eliminate the need for maintaining expensive computing facilities by companies and institutes. However, few efforts are taken to address the issue of performance evaluation in cloud environments. In this paper, we present a configurable performance evaluation analysis and evaluation framework, which is aiming to provide users and researchers an easy-to-use toolkit to evaluate their cloud system’s runtime performance, or compare the performance when different resource management policy and task scheduling algorithms are taken into account. Extensive experiments are conducted to investigate the effectiveness of the system implementation, and the results indicate that its configurable feature is of significantly usefulness when conducting performance comparing under different scenarios.