[MLS-C01] [IMPL and OPs] Introduction

Build solutions

To optimize the performance of your models, you can use Automatic Model Tuning to find the hyperparameters that will give you the optimal performance
You can use SageMaker hosting to automatically scale your model to the performance needed for your model

SageMaker hosting allows you to configure your endpoints to elastically scale your endpoint instances
Use two or more endpoint instances in different availability zones to make your endpoints highly available

SageMaker hosting automatically scales your endpoint instances to the performance needed for your application through Application Auto Scaling
You can manually adjust the number of instances and the instance type without suffering downtime by modifying your endpoint configuration
SageMaker scales to large numbers of transactions per second. The actual scale depends on your deployed model and the number and type of instances your model endpoints run on

SageMaker automatic scaling endpoint instances scale out and automatically spread instances across multiple availability zones
This gives you availability zone level fault tolerance and protects from an individual instance failure

Select from several AWS Services to implement the appropriate Machine Learning solution for your business problem

SageMaker model artifacts and other system artifacts are encrypted in transit and at rest
Requests to your endpoint API and can be mode over a secure (SSL) connection
Assign IAM roles to your model instance to provide permission to access resources on your behalf for deployment
Use encrypted S3 buckets for model artifacts and data
Pass a KMS key to your endpoints to encrypt their attached ML storage volume

Clients send HTTPS requests to your endpoint to obtain inferences
Can deploy multiple variants of a model to the same HTTPS endpoint (test variations of your model in production)
Can also configure a production variant to achieve auto scaling
You can modify your endpoints without suffering downtime by modifying your endpoint configuration

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