WebOps #2
You should probably store your images/uploads on S3. Are you going to build a cluster of synchronized fileservers? Probably not.
Unsolved question: How do you properly backup the S3 files? Backups would need to be stored somewhere off S3, in case your S3 account mysteriously disappears. And a simple copy/replace/rsync wouldn’t do it, what if all your files were truncated to zero length? You’d want some sort of an incremental backups that takes a snapshot of how your S3 account looks at a given point in time. You also need to be able to do this without doing a zillion GET/PUT/LIST requests, those are expensive.
