SageMaker Notebook CDK Template

This project provides a CDK template that provisions a SageMaker notebook instance with configurable properties.

Architecture

The template creates the following AWS resources:

• SageMaker Notebook Instance: A managed Jupyter notebook environment
• IAM Role: Service role for the notebook instance with appropriate permissions

Features

• Configurable Instance Type: Choose the compute capacity (default: ml.t3.medium)
• Configurable Storage: Set the volume size in GB (default: 5 GB)
• Secure Setup: Includes proper IAM role with SageMaker permissions
• Cost Optimized: Uses inexpensive instance types by default

Configuration

The stack accepts the following properties:

interface SageMakerNotebookStackProps {
  instanceType?: string;     // Default: 'ml.t3.medium'
  volumeSizeInGB?: number;   // Default: 5
}

Example Usage

import { SageMakerNotebookStack } from './lib/sagemaker-notebook-stack';

// Default configuration
new SageMakerNotebookStack(app, 'MyNotebookStack');

// Custom configuration
new SageMakerNotebookStack(app, 'MyNotebookStack', {
  instanceType: 'ml.t3.large',
  volumeSizeInGB: 20
});

Available Instance Types

Common SageMaker notebook instance types:

• ml.t3.medium (2 vCPU, 4 GB RAM) - Default, cost-effective
• ml.t3.large (2 vCPU, 8 GB RAM)
• ml.t3.xlarge (4 vCPU, 16 GB RAM)
• ml.m5.large (2 vCPU, 8 GB RAM)
• ml.m5.xlarge (4 vCPU, 16 GB RAM)

Setup and Deployment

Prerequisites

• Node.js 18+ installed
• AWS CLI configured with appropriate credentials
• CDK bootstrapped in your target AWS account/region

Installation

npm install

Build

npm run build

Test

npm test

Deploy

# Synthesize CloudFormation template
npm run synth

# Deploy the stack
npm run deploy

# View diff against deployed stack
npm run diff

# Destroy the stack
npm run destroy

Testing

The project includes comprehensive tests that verify:

• ✅ SageMaker notebook instance creation with default configuration
• ✅ SageMaker notebook instance creation with custom configuration
• ✅ IAM role creation with proper trust policy and permissions
• ✅ Correct resource counts (exactly one instance and one role)
• ✅ Proper CloudFormation outputs
• ✅ Instance type configurability
• ✅ Volume size configurability
• ✅ Proper resource references between components

Run tests with:

npm test

Outputs

The stack provides the following outputs:

• NotebookInstanceArn: ARN of the created SageMaker notebook instance
• NotebookInstanceName: Name of the notebook instance
• NotebookRoleArn: ARN of the IAM role used by the notebook instance

Cost Considerations

• Default instance type (ml.t3.medium) is cost-effective for development and light workloads
• Default storage volume (5 GB) is minimal to keep costs low
• Remember to stop or delete the notebook instance when not in use to avoid charges

Security

The IAM role created for the notebook instance includes:

• Trust policy allowing SageMaker service to assume the role
• AmazonSageMakerFullAccess managed policy for SageMaker operations

For production use, consider implementing more restrictive IAM policies based on your specific requirements.