DynamoDB is used *everywhere* in AMZN Retail, this is absolutely not surprising. Plus the vast majority of internal Services are using EC2 in the form of Apollo/ECS. So OP probably hit some parts of the site that are hosted in us-west-2. For all I know they started routing all requests for us-east-1 traffic to other DCs, figuring latency is a fine trade-off for availability