用 Amazon SageMaker HyperPod 实现多团队共享 GPU 集群的隔离与公平
原标题：Share GPU clusters across teams with isolation and fairness using Amazon SageMaker HyperPod
AI 摘要
AWS 发布了一篇技术博客，介绍如何在 Amazon SageMaker HyperPod 上构建多租户架构，让多个团队共享同一 GPU 集群。该方案结合 AWS IAM Identity Center 做集中认证、按团队划分 SageMaker AI 域、用 Kubernetes 命名空间隔离工作负载，并通过 HyperPod Task Governance 实现公平资源分配和按命名空间的成本分摊。目标是解决共享 GPU 时的资源失控、隔离薄弱和成本归属不清等问题。
正文节选
Share GPU clusters across teams with isolation and fairness using Amazon SageMaker HyperPod Multiple teams within the same company increasingly need shared access to expensive GPU clusters for their generative AI operations, while maintaining isolation boundaries, resource fairness, and operational independence. Consider a data science team training large language models, a computer vision group running inference workloads, and a research team experimenting with new model architectures. All of t