Loading...
Loading...

Go to the content (press return)

Adaptive request scheduling for the I/O forwarding layer using reinforcement learning

Author
Bez, J.; Zanon, F.; Nou, R.; Miranda, A.; Cortes, A.; Navaux, P.
Type of activity
Journal article
Journal
Future generation computer systems
Date of publication
2020-11
Volume
112
First page
1156
Last page
1169
DOI
10.1016/j.future.2020.05.005
Project funding
High performance computing VII
Models de Programacio i Entorns d'eXecució PARal.lels
Repository
http://hdl.handle.net/2117/331392 Open in new window
URL
https://www.sciencedirect.com/science/article/abs/pii/S0167739X20300856 Open in new window
Abstract
In this paper, we propose an approach to adapt the I/O forwarding layer of HPC systems to applications’ access patterns. I/O optimization techniques can improve performance for the access patterns they were designed to target, but they often decrease performance for others. Furthermore, these techniques usually depend on the precise tune of their parameters, which commonly falls back to the users. Instead, we propose to do it dynamically at runtime based on the I/O workload observed by the sys...
Citation
Bez, J. [et al.]. Adaptive request scheduling for the I/O forwarding layer using reinforcement learning. "Future generation computer systems", Novembre 2020, vol. 112, p. 1156-1169.
Keywords
Auto-tuning, High performance I/O, I/O forwarding, I/O scheduling, Parallel I/O, Reinforcement learning
Group of research
CAP - High Performace Computing Group

Participants

  • Bez, Jean Luca  (author)
  • Zanon Boito, Francieli  (author)
  • Nou, Ramon  (author)
  • Miranda Bueno, Alberto  (author)
  • Cortes Rossello, Antonio  (author)
  • Navaux, Philippe O.A.  (author)