Bridging the gap between profiling and monitoring in HPC systems with dynamically reconfigurable fine-grain data collection - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Reports Year : 2022

Bridging the gap between profiling and monitoring in HPC systems with dynamically reconfigurable fine-grain data collection

Abstract

Job-aware monitoring of HPC systems is a middle ground between system monitoring and profiling as it collects performance data at the scale of a job while keeping the overhead at a reasonable amount to allow for production usage. This paper presents Colmet, a job-level monitoring system capable of collecting metrics at different sampling period at the same time. The sampling periods can be dynamically reconfigured to suit the current need of the users up to the millisecond. We these features, Colmet can be used for usual monitoring purposes and for profiling purposes. Fine grain and reconfigurable data can also be used in systems that make decisions based on performance data. Our system is at least as efficient as state-of-the-art monitoring systems as we show in the performance analysis results.
Fichier principal
Vignette du fichier
report_internship.pdf (273.11 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03773464 , version 1 (21-09-2022)

Identifiers

  • HAL Id : hal-03773464 , version 1

Cite

Ilya Meignan--Masson. Bridging the gap between profiling and monitoring in HPC systems with dynamically reconfigurable fine-grain data collection. [0] Université Grenoble Alpes. 2022. ⟨hal-03773464⟩
30 View
45 Download

Share

Gmail Facebook X LinkedIn More