Resumen de Methodology to evaluate performance of the I/O systems on High Performance Computers
The increase in computational power of processing units and the complexity of scientific applications which use high performance computing require more efficient Input/Output (I/O) systems. To use the I/O systems efficiently it is necessary to know its performance capacity to determine if it fulfills I/O requirements of the application. Evaluating the performance capacity of the I/O system is difficult due to the diversity of architectures and the complexity of its software stack. Furthermore, parallel scientific applications will have different behavior depending on their access patterns. Then, it is necessary to have some method to evaluate the I/O system performance taking into account the applications access patterns. We propose a methodology to evaluate performance of the Input/Output systems on High Performance Computers based on an I/O model for the parallel scientific applications [1] [2] [3]. The proposed methodology is composed of three stages: Characterization, I/O analysis and Evaluation.
