[MLS-C01] [IMPL and OPs] Security

SageMaker Instances

Your SageMaker infrastructure uses EC2 instances dedicated for your use
Can map your SageMaker resources to VPC so you can use your network controls
Control access to your jupyter notebooks and your hasted models through IAM
Can only access your SageMaker resources from within your VPC using your VPC Endpoints (Private Connectivity)
Encrypt your data at rest and in flight from your datasets on S3 to your notebooks and through to your hosted endpoints
You can use lifecycle configurations to harden the OS of your SageMaker EC2 instances or install agents
SageMaker is integrated with CloudWatch and CloudTrail for logging training job and hosted model activity as well as API calls

Can map your training, tuning, and hosted model endpoint instances to VPC so you can use your network controls
Can restrict your training, tuning, and endpoint instances to resources within your VPC using your VPC Endpoints (Private Connectivity)

本文由 Oscaner 创作, 采用知识共享署名4.0 国际许可协议进行许可
本站文章除注明转载/出处外, 均为本站原创或翻译, 转载前请务必署名