This document discusses strategies for building applications that can scale on AWS. It recommends architecting for loose coupling between components to allow independent scaling. Using services like SQS as buffers between components decouples them. Auto scaling helps applications automatically scale compute resources up and down based on demand. Elastic Load Balancing distributes traffic across multiple instances. Session state should reside outside of scaled components, in a performant and scalable data store like DynamoDB. Data tier scaling may involve strategies like sharding or using managed databases like DynamoDB or Redshift that scale horizontally. AWS services like ELB, Auto Scaling, SQS, and databases help applications achieve web-scale.