What is the MOST operationally efficient solution to deploy a patched open-source library for all three Lambda functions?
Create a custom AWS CloudFormation public registry extension. Reference a GitHub repository that hosts the open-source modules in the extension. Configure CloudFormation to scan the repository once each day. Write an AWS Serverless Application Model (AWS SAM) template to redeploy the three Lambda functions upon a scan notification change.
Create an Amazon CloudFront distribution with an Amazon S3 bucket as the origin. Upload the patched modules to Amazon S3 when needed. Modify each Lambda function to download the patched modules from the CloudFront distribution during the cold start.
Launch an Amazon EC2 instance. Host a private open-source module registry on the EC2 instance. Upload the modified open-source modules to the private registry when needed. Modify each Lambda function deployment script to download the modules from the private registry. Redeploy the three new Lambda functions.
Create a Lambda layer with the open-source modules. Modify all three Lambda functions to depend on the layer. Remove the open-source modules from each Lambda function. Patch the Lambda layer with the modified open-source modules when needed. Update the Lambda functions to reference the new layer version.
Explanations:
AWS CloudFormation is used for infrastructure management, not for the operational efficiency of patching Lambda functions with open-source modules. Creating a CloudFormation public registry extension and relying on a GitHub repository would add unnecessary complexity and is not the most efficient way to update Lambda functions.
Using Amazon CloudFront and S3 to host and download patched modules at Lambda cold start introduces performance overhead. This approach requires managing the download process within each Lambda function and does not provide a streamlined, efficient method for dependency management.
Hosting a private registry on an EC2 instance introduces the complexity of maintaining the EC2 instance and requires custom code to handle the downloading of the modules. This approach does not optimize for operational efficiency, especially compared to simpler solutions like Lambda layers.
Using a Lambda layer to manage shared dependencies like open-source modules is the most efficient and operationally sound solution. Lambda layers allow for versioned dependencies that can be reused across multiple functions, and patching the layer automatically updates all dependent Lambda functions.
I deduce that the answer is:
Create a Lambda layer with the open-source modules. Modify all three Lambda functions to depend on the layer. Remove the open-source modules from each Lambda function. Patch the Lambda layer with the modified open-source modules when needed. Update the Lambda functions to reference the new layer version.