Amazon, while not quite as fast as Rackspace, is accessed in a similar way. We could use a Web-based upload or download, or Amazon's API. The API in turn, offered myriad language choices, including Ruby, .Net/C#, Java, and many more — as well as third party client access tools like CyberDuck. Amazon adds BitTorrent access for download as well. And like Rackspace, S3 has a limitation of 5GB per object.
Amazon's pricing is similar to Rackspace's, with charges for storage (first 50Tbytes at 15cents per gigabyte, data transfers and requests (1 cent per 1,000 requests.) An added twist is that Amazon offers slightly less expensive storage depending on where your Amazon S3 bucket is located.
Unlike Rackspace, Amazon's S3 implements access control lists on files. This comes in the form of read, write, and "full control" metadata on objects. As we went to press, Amazon released new controls over buckets (containers) that ostensibly secure the container and all of the files inside.
S3 can also do a query string authentication, so that requests to manipulate files can be made into a URL directly to an S3 object. Overall access is performed through an Access Key ID and a Secret Key, generated through the Amazon Web interface control panel. Multiple keys can be generated at a time for subsequent use by a primary account holder for use by others, including application use.
Natively, the Amazon S3 storage system is surprisingly flexible, and can be used for the public/private cloud computing components aforementioned, as well as Web apps, backups, temporary storage use, file distribution, plus distribution via BitTorrent.