A scalable synthetic traffic model of Graph500 for computer networks analysis
Ver/ Abrir
Identificadores
URI: http://hdl.handle.net/10902/13658DOI: 10.1002/cpe.4231
ISSN: 1532-0626
ISSN: 1532-0634
Registro completo
Mostrar el registro completo DCAutoría
Fuentes Saez, Pablo; Benito Hoz, Mariano; Vallejo Gutiérrez, Enrique; Bosque Orero, José Luis; Beivide Palacio, Ramón; Anghel, Andreea; Rodríguez, Germán; Gusat, Mitch; Minkenberg, Cyriel; Valero, MateoFecha
2017-12Derechos
© John Wiley & Sons. This is the peer reviewed version of the following article: Pablo Fuentes, Mariano Benito, Enrique Vallejo, José Luis Bosque, Ramón Beivide, Andreea Anghel, Germán Rodríguez, Mitch Gusat, Cyriel Minkenberg, Mateo Valero: A scalable synthetic traffic model of Graph500 for computer networks analysis. Concurrency and Computation: Practice and Experience (2017), Vol.19, n.24, which has been published in final form at https://doi.org/10.1002/cpe.4231. This article may be used for non-commercial purposes in accordance with Wiley Terms and Conditions for Self-Archiving.
Publicado en
Concurrency and Computation: Practice and Experience (2017), vol.19, n.24
Editorial
John Wiley & Sons
Resumen/Abstract
The Graph500 benchmark attempts to steer the design of High-Performance Computing sys-tems to maximize the performance under memory-constricted application workloads. A realisticsimulation of such benchmarks for architectural research is challenging due to size and detail lim-itations. By contrast, synthetic traffic workloads constitute one of the least resource-consumingmethods to evaluate the performance. In this work, we provide a simulation tool for networkarchitects that need to evaluate the suitability of their interconnect for BigData applications. Ourdevelopment is a low computation- and memory-demanding synthetic traffic model that emu-lates the behavior of the Graph500 communications and is publicly available in an open-sourcenetwork simulator. The characterization of network traffic is inferred from a profile of several exe-cutions of the benchmark with different input parameters. We verify the validity of the equationsin our model against an execution of the benchmark with a different set of parameters. Further-more, we identify the impact of the node computation capabilities and network characteristics inthe execution time of the model in a Dragonfly network.
Colecciones a las que pertenece
- D30 Artículos [72]
- D30 Proyectos de Investigación [83]