Make CloudFront resize images "on the fly" via lambda@edge, cache it and persists it in S3. Utilises Sharp for image transformations.

Schema

Illustration & inspiration from this blog post

Since this relies on compiled binaries for libvips, the package & deploy is a bit cumbersome.

rm -rf node_modules
docker run -v "$PWD":/var/task lambci/lambda:build-nodejs6.10 /bin/bash -c 'npm install -g serverless && npm install && sls package'
sls deploy -p .serverless

• Functions have to reside in us-east-1
• Node 6.10 only
• No environment variables
• Viewer-* functions are limited to 5 seconds execution and 128 MB RAM See here
• Response-* functions normal Lambda limits
• CloudFront deployment takes about 20 minutes. Hence, an update of functions takes the same time
• CloudWatch logs appear at the closest option to the edge location of a request

• Tests! Given we have to wait 20 minutes for a deployment, a "production" failure is quite costly

• Add dynamic configuration file to work around the missing environment in lambda@edge
• Make the resizing more resilient (limit to original image dimensions, allow something like x300 to scale one dimension dynamically)
• Encode dimensions in path to get rid of the viewer-request function
• Expose more Sharp features
• On-Demand external image source
  • e.g. fetch original image from mydomain.com and persist it in S3
• Add Kinsesis stream to aggregate all CloudWatch logs from functions across all edge locations