SkyNomad: On Using Multi-Region Spot Instances to Minimize AI Batch Job Cost
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866914245340299264 |
|---|---|
| author | Li, Zhifei Xia, Tian Mao, Ziming Zhou, Zihan Jackson, Ethan J. Kerney, Jamison Wu, Zhanghao Mishra, Pratik Xu, Yi Qiao, Yifan Shenker, Scott Stoica, Ion |
| author_facet | Li, Zhifei Xia, Tian Mao, Ziming Zhou, Zihan Jackson, Ethan J. Kerney, Jamison Wu, Zhanghao Mishra, Pratik Xu, Yi Qiao, Yifan Shenker, Scott Stoica, Ion |
| contents | AI batch jobs such as model training, inference pipelines, and data analytics require substantial GPU resources and often need to finish before a deadline. Spot instances offer 3-10x lower cost than on-demand instances, but their unpredictable availability makes meeting deadlines difficult. Existing systems either rely solely on spot instances and risk deadline violations, or operate in simplified single-region settings. These approaches overlook substantial spatial and temporal heterogeneity in spot availability, lifetimes, and prices. We show that exploiting such heterogeneity to access more spot capacity is the key to reduce the job execution cost. We present SkyNomad, a multi-region scheduling system that maximizes spot usage and minimizes cost while guaranteeing deadlines. SkyNomad uses lightweight probing to estimate availability, predicts spot lifetimes, accounts for migration cost, and unifies regional characteristics and deadline pressure into a monetary cost model that guides scheduling decisions. Our evaluation shows that SkyNomad achieves 1.25-3.96x cost savings in real cloud deployments and performs within 10% cost differences of an optimal policy in simulation, while consistently meeting deadlines. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_06520 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | SkyNomad: On Using Multi-Region Spot Instances to Minimize AI Batch Job Cost Li, Zhifei Xia, Tian Mao, Ziming Zhou, Zihan Jackson, Ethan J. Kerney, Jamison Wu, Zhanghao Mishra, Pratik Xu, Yi Qiao, Yifan Shenker, Scott Stoica, Ion Distributed, Parallel, and Cluster Computing C.2.4 AI batch jobs such as model training, inference pipelines, and data analytics require substantial GPU resources and often need to finish before a deadline. Spot instances offer 3-10x lower cost than on-demand instances, but their unpredictable availability makes meeting deadlines difficult. Existing systems either rely solely on spot instances and risk deadline violations, or operate in simplified single-region settings. These approaches overlook substantial spatial and temporal heterogeneity in spot availability, lifetimes, and prices. We show that exploiting such heterogeneity to access more spot capacity is the key to reduce the job execution cost. We present SkyNomad, a multi-region scheduling system that maximizes spot usage and minimizes cost while guaranteeing deadlines. SkyNomad uses lightweight probing to estimate availability, predicts spot lifetimes, accounts for migration cost, and unifies regional characteristics and deadline pressure into a monetary cost model that guides scheduling decisions. Our evaluation shows that SkyNomad achieves 1.25-3.96x cost savings in real cloud deployments and performs within 10% cost differences of an optimal policy in simulation, while consistently meeting deadlines. |
| title | SkyNomad: On Using Multi-Region Spot Instances to Minimize AI Batch Job Cost |
| topic | Distributed, Parallel, and Cluster Computing C.2.4 |
| url | https://arxiv.org/abs/2601.06520 |