Navigating the Migration from SageMaker Studio Classic to the Modern Experience

A step-by-step guide to upgrade to the new SageMaker Studio

Key Highlights

End of Support for Classic: SageMaker Studio Classic will no longer receive critical updates after December 2024, emphasizing the need to migrate.
Smooth Transition: Use test domains and AWS CLI tools to ensure a safe and seamless migration of your notebooks, data, and custom configurations.
Enhanced Features: The new Studio supports JupyterLab 4 with improved performance, modern UI, and better integration with AWS services.

Overview of the Migration Process

Migration from Amazon SageMaker Studio Classic to the modern SageMaker Studio involves several planned steps. The new Studio environment is designed to provide improved functionality, updated user experience, and enhanced support for machine learning workloads. Below we outline the process to create a safe, effective migration strategy.

1. Preparatory Assessment and Environment Review

Evaluate Your Current Setup

Before beginning the migration, perform a comprehensive review of your existing SageMaker Studio Classic environment. This includes identifying custom images, configurations, lifecycle scripts, dependencies, and data flows. By understanding what components require migration, you can plan the necessary adjustments for the new Studio experience.

Key tasks during the assessment phase include:

Cataloging custom images and lifecycle configuration scripts.
Reviewing network settings and VPC configurations.
Identifying any Amazon S3 buckets, EFS volumes, or data storage requirements tied to Studio Classic.
Documenting specific notebook dependencies, packages, and customizations.

2. Enable the New Studio Experience

Switching Default Settings and Domain Updates

Accounts created before November 30, 2023, may have SageMaker Studio Classic set as the default experience. To access the latest features, switch to the modern Studio using either the Amazon SageMaker AI console or the AWS Command Line Interface (AWS CLI). Use the available CLI commands or interface options to update your domain settings.

For example, one approach with AWS CLI is:


# Replace placeholders with your specific domain ID and region
aws sagemaker update-domain --domain-id your-domain-id --region your-region --default-user-settings '{ "StudioWebPortal": "ENABLED", "DefaultLandingUri": "app:JupyterServer:" }'

This command enables the modern SageMaker Studio experience whereby your user settings and access can be reconfigured to take advantage of JupyterLab 4 and modern UI enhancements.

3. Create a Test Domain and Validate the Migration

Setting Up a Sandbox Environment

It is recommended to create a test domain in the new Studio that mirrors the configurations of your current Studio Classic environment. This test domain acts as a sandbox where you can verify compatibility with your organization’s network, security settings, and custom configurations without impacting production workloads.

In the test domain setup, ensure the following:

The domain replicates the network settings and VPC configurations of the original Studio Classic.
All user preferences, custom images, and lifecycle scripts are set up similarly.
Custom data connections, including access to S3 or EFS, are accurately recreated for testing.

4. Migrate the User Interface (UI) and Custom Configurations

Seamless Integration of the UI

The migration involves transitioning the user interface from Studio Classic to the new Studio. The UI of the old environment—based on JupyterLab 3—will be updated to align with the modern features available on JupyterLab 4. This ensures both backward compatibility for your legacy notebooks as well as access to new functionalities.

Steps include:

Utilizing the AWS CLI or SageMaker Console to update the domain's UI configuration.
Verifying that all network and security settings are in place, especially for organizations with specific compliance needs.
Testing functionalities such as notebook access, terminal usage, and file system connectivity in the new UI.

5. Handling Data, Custom Images, and Lifecycle Configurations (Optional)

Data Migration Considerations

If your workload in Studio Classic involves custom images or dedicated lifecycle configuration scripts, you may need to migrate these components manually. Although the core migration handles the UI and primary environment, extra steps might be required for:

Custom Images: Recreate these images in the new Studio to ensure continuity in the machine learning environment. Validate that the packages and dependencies are the same.
Lifecycle Scripts: Replicate or adjust existing lifecycle configuration scripts to work with the updated environment.
Data Migration: Studio Classic relies on a shared Amazon EFS volume while the new Studio typically uses individual EBS volumes for each compute space. During migration, consider:
- Determining if data is shared between applications.
- Transferring data from the legacy EFS to corresponding EBS volumes or migrating to Amazon S3 buckets as needed.

Comparative Table: SageMaker Studio Classic vs. Modern Studio

Feature/Aspect	SageMaker Studio Classic	Modern SageMaker Studio
Default JupyterLab Version	JupyterLab 3	JupyterLab 4
UI and User Experience	Classic interface with limited recent updates	Enhanced modern UI with new capabilities
Data Storage	Shared Amazon EFS volume	Individual EBS volumes per compute space
Support & Updates	End of support and critical fixes post-December 2024	Active updates and feature enhancements
Custom Configurations	Requires manual replication of custom images and lifecycle scripts	Migration supports easier transfer with updated configurations
Default Experience	Accounts created before November 30, 2023	All new accounts or migrated instances

Detailed Migration Steps for Implementation

Step-by-Step Breakdown

Step 1: Preparation

Audit your current Studio Classic environment: List all notebooks, configurations, custom images, lifecycle scripts, and data storages.
Identify non-replicable dependencies and check for any third-party integrations.
Document the configurations and network settings to mirror in the new environment.

Step 2: Setting Up the New Studio Domain

Use the AWS CLI or SageMaker AI console to enable the new Studio experience.
Create a test domain with the same VPC and network configurations to ensure reliability.
Verify IAM role permissions are updated to support modern Studio operations, including required policies (e.g., access to S3 and EBS/EFS resources).

Step 3: Migrating the UI

Update the domain settings via the AWS CLI by switching the default experience from Studio Classic to the modern Studio.
Ensure that the updated configuration applies to all users by confirming deployment through the SageMaker AI console.
Test network connectivity and user access post-UI migration to confirm that all functionalities are intact.

Step 4: Transferring Data and Custom Configurations

Migrate user data: If using shared storage (EFS) in Studio Classic, consider transferring essential files to the individual EBS volumes associated with the modern Studio setup, or alternatively, use Amazon S3 for file transfer.
Recreate custom images and lifecycle configurations: Manually recreate or update these resources within the new Studio to mirror the existing environment.
Validate that custom scripts run successfully in the new environment and adjust configuration parameters if needed.

Step 5: Final Testing and Operational Validation

Perform a thorough testing of the new domain covering notebooks, terminal interfaces, and file system mappings.
Review integration points with other AWS services to ensure seamless interoperability.
After successful testing in the sandbox, plan the full production migration ensuring minimal downtime.

After following these detailed steps, your applications and workloads will transition smoothly from SageMaker Studio Classic to the modern AWS SageMaker Studio environment. This ensures continued access to critical fixes, enhancements, and support for advanced machine learning projects.

Additional Considerations and Best Practices

Security and Permissions

Confirm that execution roles and IAM policies are configured correctly in the modern Studio. Ensure that all permissions for accessing Amazon S3 buckets, EBS volumes, and other resources are aligned with your organization’s security guidelines.

Performance and Resource Management

The modern SageMaker Studio uses individual EBS volumes for data, offering performance improvements and isolation compared to the shared EFS in Classic. Monitor resource usage post-migration and adjust configurations to optimize performance.

Documentation and AWS Resources

Rely on official AWS documentation for the latest updates on the migration process and best practices. Regularly review AWS announcements for any changes or enhancements in migration procedures.