Metraipolis › archive › dc_2ae254aa85b1
Argus: A Real-EKS Study of When Predicting Spot Interruptions Beats Simple Checkpointing — arXiv:2609.39067
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.39067
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T13:15:00.774Z
- SHA-256 of the archived text
c978c4ec02193e735c84b7121a0b10d5e2895586c4a7e50a40264163c2a77b90- Stable URL of this page
https://metraipolis.com/documento/dc_2ae254aa85b1
How to cite
Argus: A Real-EKS Study of When Predicting Spot Interruptions Beats Simple Checkpointing — arXiv:2609.39067 — arXiv (en), https://arxiv.org/abs/2609.39067 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2ae254aa85b1 (Metraipolis archive dc_2ae254aa85b1, sha256 c978c4ec02193e735c84b7121a0b10d5e2895586c4a7e50a40264163c2a77b90)
BibTeX
@misc{dc2ae254aa85b1,
title = {Argus: A Real-EKS Study of When Predicting Spot Interruptions Beats Simple Checkpointing — arXiv:2609.39067},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.39067},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2ae254aa85b1; sha256 c978c4ec02193e735c84b7121a0b10d5e2895586c4a7e50a40264163c2a77b90},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.39067. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Elastic Compute Cloud (EC2) Spot is 60% to 90% cheaper than On-Demand but can be reclaimed on just a 2-minute notice; for expensive multi-node training this loss can be severe, with one reclaim costing hours of synchronous progress. We build Argus, a Kubernetes operator, and ask empirically, on a CIFAR-10 testbed, when predicting interruptions beats simple checkpointing. Argus on real EKS survives a real Spot drain with a graceful SIGTERM checkpoint, resuming from epoch 8 and losing only the in-progress epoch. Alongside, we further find that in an 80-trial benchmark, the reactive-on-notice degrades toward no protection once interruption outpaces the fixed 2-minute notice, and predictive wasted compute is driven to zero, but with an oversized fixed lead it over-migrates so severely that at the fastest rate only one of five runs completes, while periodic is a strong ML-free baseline. A lead-time sweep turns the lead prediction into a guideline where a small lead suffices for zero waste, but excess lead is wasteful. The predictor built is advisory (a proxy label); real interruption labels and large-model-scale validation are future work. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-10-01 in cs.DC). Categorie arXiv: Machine Learning (cs.LG) ; Distributed, Parallel, and Cluster Computing (cs.DC). Pagina del lavoro: https://arxiv.org/abs/2609.39067 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-07T12:46:39.752Z dalla pagina https://arxiv.org/catchup/cs.DC/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).