Increase efficiency using Lambda layers with AWS SAM

Using AWS SAM, you can include layers in your serverless applications. AWS Lambda layers allow you to extract code from a Lambda function into a Lambda layer which can then be used across several Lambda functions. Doing this allows you to reduce the size of your deployment packages, separate core function logic from dependencies, and share dependencies across multiple functions. For more information about layers, see Lambda layers in the AWS Lambda Developer Guide.

This topic provides information about the following:

For information about building custom layers, see Building Lambda layers in AWS SAM.

Including layers in your application

To include layers in your application, use the Layers property of the AWS::Serverless::Function resource type.

Following is an example AWS SAM template with a Lambda function that includes a layer:

ServerlessFunction:
  Type: AWS::Serverless::Function
  Properties:
    CodeUri: .
    Handler: my_handler
    Runtime: Python3.7
    Layers:
        - <LayerVersion ARN>

How layers are cached locally

When you invoke your function using one of the sam local commands, the layers package of your function is downloaded and cached on your local host.

The following table shows the default cache directory locations for different operating systems.

After the package is cached, the AWS SAM CLI overlays the layers onto a Docker image that's used to invoke your function. The AWS SAM CLI generates the names of the images it builds, as well as the LayerVersions that are held in the cache. You can find more details about the schema in the following sections.

To inspect the overlaid layers, execute the following command to start a bash session in the image that you want to inspect:

docker run -it --entrypoint=/bin/bash samcli/lambda:<Tag following the schema outlined in Docker Image Tag Schema> -i

Layer Caching Directory name schema

Given a LayerVersionArn that's defined in your template, the AWS SAM CLI extracts the LayerName and Version from the ARN. It creates a directory to place the layer contents in named LayerName-Version-<first 10 characters of sha256 of ARN>.

Example:

ARN = arn:aws:lambda:us-west-2:111111111111:layer:myLayer:1
Directory name = myLayer-1-926eeb5ff1

Docker Images tag schema

To compute the unique layers hash, combine all unique layer names with a delimiter of '-', take the SHA256 hash, and then take the first 10 characters.

Example:

ServerlessFunction:
  Type: AWS::Serverless::Function
  Properties:
    CodeUri: .
    Handler: my_handler
    Runtime: Python3.7
    Layers:
        - arn:aws:lambda:us-west-2:111111111111:layer:myLayer:1
        - arn:aws:lambda:us-west-2:111111111111:layer:mySecondLayer:1

Unique names are computed the same as the Layer Caching Directory name schema:

arn:aws:lambda:us-west-2:111111111111:layer:myLayer:1 = myLayer-1-926eeb5ff1
arn:aws:lambda:us-west-2:111111111111:layer:mySecondLayer:1 = mySecondLayer-1-6bc1022bdf

To compute the unique layers hash, combine all unique layer names with a delimiter of '-', take the sha256 hash, and then take the first 25 characters:

myLayer-1-926eeb5ff1-mySecondLayer-1-6bc1022bdf = 2dd7ac5ffb30d515926aef

Then combine this value with the function's runtime and architecture, with a delimiter of '-':

python3.7-x86_64-2dd7ac5ffb30d515926aefffd