Lambda Layers

Updating the function was kind of slow! We can use Layers to help there.

Layers are just a .zip archive of extra STUFF (files) you can have Lambda add to your function automatically. This lets you keep the function itself smaller, and faster to update.

They require a certain directory structure, based on the runtime.

For example, for a NodeJS runtime, it needs files to be in a nodejs directory:

1# We need a certain directory structure...
2mkdir -p layer/nodejs
3cp -r node_modules package* layer/nodejs/
4 
5# Zip it up
6cd layer # Note we zip it from this `layer` directory
7zip -r9q layer.zip .

Once it's zipped up (note the file paths used), we can create the layer, and update the function to use it:

 1# Create the layer
 2aws --profile cloudcasts lambda publish-layer-version \
 3    --layer-name "foo_nodejs_modules" \
 4    --description "AWS S3 Node Modules" \
 5    --zip-file fileb://layer.zip \
 6    --compatible-runtimes nodejs14.x
 7 
 8# Update the function
 9aws --profile cloudcasts lambda update-function-configuration \
10    --function-name foo_func_nodejs \
11    --layers "arn:aws:lambda:us-east-2:601364042938:layer:foo_nodejs_modules:2"

The function is now using the layer.

Let's repackage our function (it's now just the index.js file) and re-upload it so the function is back to being tiny:

 1# Create a new .zip of the app (just the `index.js` file!)
 2zip -r9 index.zip index.js
 3 
 4# Update the function
 5aws --profile cloudcasts lambda update-function-code \
 6    --function-name foo_func_nodejs \
 7    --zip-file fileb://index.zip \
 8    --publish
 9 
10# Invoke the function
11aws --profile cloudcasts lambda invoke \
12    --function-name foo_func_nodejs \
13    --invocation-type RequestResponse \
14    output.json
15 
16cat output.json

And it (still) worked! Our function is tiny, the layer has the larger stuff. This makes updating the function must quicker.