by Luigi Patruno, Founder of MLinProduction.com
TODO Recorded AWS Webinar
- Deployment Strategies for Machine Learning Services
- Single Deployment
- Silent Intelligence
- Controlled Rollout
- Flighting
- Setup
- Upload necessary datasets to S3 bucket.
- Train multiple predictive models.
- Deply models as persistent API endpoints.
- View
Endpoint
andEndpointConfig
details.
- Blue Green Deployment
- Create an Endpoint with a single Production Variant.
- Create a new endpoint configuration, using the same production variants for the existing live model and for the new model.
- Update the existing live endpoint with the new endpoint configuration. Amazon SageMaker creates the required infrastructure for the new production variant and updates the weights without any downtime.
- Switch traffic to the new model through an API call.
- Create a new endpoint configuration with only the new production variant and apply it to the endpoint.
- Canary Deployment
- Create an Endpoint with a single Production Variant.
- Update the existing live endpoint with the new endpoint configuration. Amazon SageMaker creates the required infrastructure for the new production variant and updates the weights without any downtime.
- Gradually move traffic from one variant to the other.
- Create a new endpoint configuration with only the new production variant and apply it to the endpoint.
- References