Title: An effort to characterise enhancements I/O of storage environments

Authors: Laercio Pioli; Victor Ströele; Mario A.R. Dantas

Addresses: Department of Computer Science, Federal University of Juiz de Fora - UFJF, Juiz de Fora, MG, Brazil ' Department of Computer Science, Federal University of Juiz de Fora - UFJF, Juiz de Fora, MG, Brazil ' Department of Computer Science, Federal University of Juiz de Fora - UFJF, Juiz de Fora, MG, Brazil

Abstract: Data management and storage are becoming challenging nowadays due to the huge amount of created, processed, and stored data. The growing gap between power processing and storage latency increases this performance disparity. Targeting reducing I/O bottleneck in storage environments, researchers are proposing interesting improvements in I/O architectures. High-Performance Computing (HPC) and Data-Intensive Scalable Computing (DISC) applications are kinds of such systems that are faced with data challenges due to the need to deal with many parameters when managing data. This study describes our characterisation model for classifying research works on I/O performance improvements for storage systems and devices that improve HPC and DISC overall application performance. We present a set of experiments using a synthetic I/O benchmark performed inside the Grid'5000. We show that the latency when performing I/O operations can undergo many variations if we take into account the presented factors evaluated in the experiments.

Keywords: I/O characterisation; I/O performance; I/O improvement; I/O model; HPC; DISC; Big Data; GRID5000; storage system; storage environments.

DOI: 10.1504/IJGUC.2023.129704

International Journal of Grid and Utility Computing, 2023 Vol.14 No.1, pp.51 - 61

Received: 11 Nov 2020
Accepted: 14 May 2021
Published online: 21 Mar 2023 *