《大数据处理实践探索》---- 基于Flask lambda 服务在aws 上 进行模型部署

In order to work around the package size limit of aws lambda. There’re three options

• 1. If compressed deployment package size > 50MB and uncompressed package size < 250MB. You can upload your package through s3. Here’s a tutorial:

https://dashbird.io/blog/exploring-lambda-limitations/

• 2. If compressed deployment package size > 50MB and uncompressed package size > 250MB, you need to deploy your stuff on AWS Sagemaker and create an endpoint to expose the model prediction function as a REST service.

• 3. Create an EC2 instance and serve the model using a flask server. The lambda function basically receive events from Kinesis and send it out directly to the flask server, Then get response from the flask server.

For option-1,

Th