I remember talking to someone who was playing with AWS stuff for the first time and they had a similar architecture, using Dynamo for a lookup store. It still seems a bit odd to me though. It's been a long time since I've worked with the S3 API, so maybe it just doesn't support the same sort of thing, but wouldn't it be nicer to just query S3 with some key and get back either the path/URL to render a link, or the content itself? Why the Dynamo intermediary? (And on the other side, if you don't need to render a link to serve the content, why not use Dynamo as the actual document store and skip S3? Storage cost?)