Somebody suggested using Amazon EC2 to provide an incremental backup interface to S3. For the likes of rsync to be truly effective, you need to have an rsync program running near the data store. It struck me, you could use an EC2 instance with free, high-speed bandwidth to S3 to do exactly that.

Using S3fuse (currently down) I’m sure you could script something so users log in to the EC2 instance and then mount their S3 bucket locally. Then rsync would work pretty effectively I reckon.

Could even be an interesting business model.