Since a bucket name is global across the S3 service (akin to a domain name), each developer account can have up to 100 buckets at any one time. Bucket names can contain only alphanumeric characters, underscores (_), periods (.), and hyphens (-). They must be between 3 and 255 characters long, and buckets with names containing uppercase characters are not accessible using the virtual hosting method.^[264]

Objects consist of object data and associated metadata. An object can hold up to 5 gigabytes of data.^[265]

A key is like a filename for an object and must be unique within a bucket. Its UTF-8 encoding must be at most 1,024 bytes long.

You use prefixes and delimiters in keys to simulate a hierarchical (folder within ­folder-­like) organization within buckets.^[266]

For both REST and SOAP requests to S3, the user metadata size associated with objects is limited to 2,000 bytes. They are structured as ­key-­value pairs.^[267]

There is an authentication and authorization system in place. You can have ­fine-­grained authorization, where you can associate permissions with specific users or with larger preset groups (the owner, everyone, or authenticated users). Permissions are read, write, or full control.^[268]

You can retrieve a .torrent file for any publicly available object by adding a ?torrent query string parameter at the end of the REST GET request for the object.^[269]

There is virtual hosting of buckets that allows one to associate your own ­non-­Amazon.com domain name with an S3 bucket.^[270]

http://s3.amazonaws.com/{bucket}/{key}

are also accessible at the following (provided that the bucket name has no uppercase characters):

http://{bucket}.s3.amazonaws.com/{key}

For example, the following:

http://s3.amazonaws.com/raymondyee/858Xtoc___.pdf

is accessible here:

http://raymondyee.s3.amazonaws.com/858Xtoc___.pdf