Amazon SageMaker Studio now supports managing environments on Amazon SageMaker HyperPod EKS clusters
AWS has made it possible to create, configure, start and stop Amazon SageMaker Spaces environments directly through Amazon SageMaker Studio. It supports JupyterLab and Code Editor environments on Amazon SageMaker HyperPod EKS clusters; an administrator must prepare the add-on and permissions.
AWS has added management of Amazon SageMaker Spaces environments on Amazon SageMaker HyperPod EKS clusters to Amazon SageMaker Studio. Developers can create, configure, start, stop and open environments on the IDE and Notebooks tab. They can now perform routine management tasks through a graphical interface; previously, they primarily used HyperPod CLI or kubectl commands.
JupyterLab and Code Editor development environments are available in the browser. A local Visual Studio Code editor can also connect via SSH-over-SSM, without managing SSH keys or opening port 22. Work remains saved on an attached Amazon Elastic Block Store volume, so stopping and restarting the environment does not cause saved data to be lost. According to AWS, startup with nodes and images prepared in advance takes approximately 30–40 seconds, compared with 5–7 minutes without this preparation.
An administrator must install the Amazon SageMaker Spaces add-on and configure access permissions for both the cluster and users. For access through the web interface, the source requires the Custom install option; for routing traffic to the environments, it describes AWS Application Load Balancer and custom DNS through Amazon Route 53. Shared environment templates, compute quotas and automatic node scaling can optionally be configured.
Why it matters
Interactive development can take place on the same infrastructure as model training and deployment. The new integration allows developers to operate these environments through a graphical interface, while administrators can control their configuration and access.
Two audiences, two different impacts
What this means
For individuals
A developer can open and manage JupyterLab or Code Editor environments from a browser without using commands for routine operations. Saved work can be preserved even when the environment is stopped.
For a business
Giving the team access to the feature requires preparing the add-on, permissions and, where needed, web access. Administrators can use templates to standardize environment configurations and quotas to manage the allocation of compute resources.
DevelopmentCheck the original
Event sources
clearly official source · 1 publisher, 0 independent. We count feeds from the same owner only once.