SageMaker HyperPod 与 Qumulo 实现多区域训练

AI 摘要

AWS 博客介绍将 Amazon SageMaker HyperPod 与 Qumulo 的 Cloud Native Qumulo（CNQ）及 Cloud Data Fabric（CDF）结合的多区域训练方案，让训练计算与数据集分处不同 AWS Region 而无需复制数据。验证中，位于 us-west-2 的 spoke 集群通过 CDF 远程读取 us-east-2 的数据，在 60ms 网络延迟下，经短暂预热后吞吐量达到与本地 hub 集群相同的 115–117 samples/sec，GPU 利用率收敛至 98–100%。该方案旨在避免跨区域复制 PB 级数据或承受持续跨区延迟的取舍。

正文节选

Multi-Region training with Amazon SageMaker HyperPod and Qumulo With Amazon SageMaker HyperPod and Qumulo, you can place training compute in one AWS Region and keep your dataset in another. Training large AI models requires massive GPU capacity, but your ideal compute resources and your training data don’t always reside in the same AWS Region. Accessing data across Regions adds network latency and transfer costs. Teams face a choice: either replicate petabytes of data across Regions, or absorb c

发布时间：2026-09-25 23:49

抓取时间：2026-09-26 00:15

来源机构：AWS