If you are following along with the patterns, you might notice that the previous pattern is a bit slow to display from the web server. When we iterated over the objects through the PHP API, there were a lot of calls being made in the background, such as SSL handshakes, IAM authorization for the calls and objects themselves, and so on.

We could improve heavily on the previous example by indexing the files that are to be uploaded. The proxy server that uploads to S3 could cache or store metadata about the file. This would ensure that the user gets a much quicker response, instead of waiting for the S3 calls to complete on each refresh of the web server. It should also be noted that without this improvement, the number of API calls is directly proportional to the number of end users hitting the web server. Amazon's S3 billing model is by API usage, so the cost goes up as the number of users and the amount of usage goes up. Management might be silently thankful for a heavily...