Assuming failure independence: are we right to be wrong? - Archive ouverte HAL Access content directly
Conference Papers Year :

Assuming failure independence: are we right to be wrong?

(1) , (2, 3) , (3, 4)
1
2
3
4

Abstract

This paper revisits the failure 1 temporal independence hypothesis which is omnipresent in the analysis of resilience methods for HPC. We explain why a previous approach is incorrect , and we propose a new method to detect failure cascades, i.e., series of non-independent consecutive failures. We use this new method to assess whether public archive failure logs contain failure cascades. Then we design and compare several cascade-aware checkpointing algorithms to quantify the maximum gain that could be obtained, and we report extensive simulation results with archive and synthetic failure logs. Altogether, there are a few logs that contain cascades, but we show that the gain that can be achieved from this knowledge is not significant. The conclusion is that we can wrongly, but safely, assume failure independence!
Fichier principal
Vignette du fichier
fts2017.pdf (597.12 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-01654639 , version 1 (04-12-2017)

Identifiers

Cite

Guillaume Aupy, Yves Robert, Frédéric Vivien. Assuming failure independence: are we right to be wrong?. FTS 2017 - 3rd International Workshop on Fault-Tolerant Systems, Sep 2017, Honolulu (HI), United States. pp.1-8, ⟨10.1109/CLUSTER.2017.24⟩. ⟨hal-01654639⟩
123 View
113 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More