Metraipolis › archive › dc_bea9a47557fb
OAK: Restart- and Age-Aware Scheduling for Distributed Machine Learning on Shared GPU Clusters — arXiv:2609.19024
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.19024
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-15T13:55:42.885Z
- SHA-256 of the archived text
dab6deff56ae8991b9e5b52411af019825b1b11f1144608ffad3c63cec38835c- Stable URL of this page
https://metraipolis.com/documento/dc_bea9a47557fb
How to cite
OAK: Restart- and Age-Aware Scheduling for Distributed Machine Learning on Shared GPU Clusters — arXiv:2609.19024 — arXiv (en), https://arxiv.org/abs/2609.19024 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bea9a47557fb (Metraipolis archive dc_bea9a47557fb, sha256 dab6deff56ae8991b9e5b52411af019825b1b11f1144608ffad3c63cec38835c)
BibTeX
@misc{dcbea9a47557fb,
title = {OAK: Restart- and Age-Aware Scheduling for Distributed Machine Learning on Shared GPU Clusters — arXiv:2609.19024},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.19024},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bea9a47557fb; sha256 dab6deff56ae8991b9e5b52411af019825b1b11f1144608ffad3c63cec38835c},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.19024. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Distributed machine learning increasingly runs on shared multi-tenant GPU clusters where contention and failures are routine. Goodput-driven schedulers maximise instantaneous throughput but treat accumulated waiting time and restart cost as second-class signals: long-waiting jobs are repeatedly deferred, and the minutes of checkpoint loading paid after each interruption are not folded back into allocation decisions. We present OAK, a per-round mixed-integer linear scheduler whose composite utility adds two first-class terms to goodput: an age key that elevates jobs with high cumulative waiting time, and a decomposed restart factor that tracks productive training time and checkpoint overhead as separately measured quantities rather than a single aggregate ratio. We evaluate OAK against four representative goodput- and fairness-driven baselines in a trace-driven simulator on a 12-GPU cluster across four data-parallel workloads with controlled Poisson failure injection, and reproduce the failure-mode improvement on a 4-V100 real-hardware cluster. Under failure-free conditions OAK matches the strongest goodput-driven baseline within 4%. Under failures at rate $\lambda = 0.1$ it reduces mean job completion time (JCT) by 33.9-58.3% and worst-case JCT by 58-76% over its goodput-driven foundation; the decomposed restart factor alone accounts for a 40-57% mean and 52-76% tail JCT improvement against an aggregate restart estimator used in prior goodput-driven schedulers. Per-round decision latency is 7.65 ms, three orders of magnitude faster than evolutionary-search alternatives at default settings. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-09-17 in cs.DC). Categorie arXiv: Distributed, Parallel, and Cluster Computing (cs.DC). Pagina del lavoro: https://arxiv.org/abs/2609.19024 — 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 2026-10-15T13:48:52.637Z dalla pagina https://arxiv.org/catchup/cs.DC/2026-09-17?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).