In S3 you have different storage classes that objects are stored in with different rules of access and costs. To start off we have standard access, this offers low latency and high throughput, is the more expensive option and is useful for big data analytics with 99.99% availability. Next up we have Standard IA, infrequent access, this class stores objects for much cheaper however it does charge an access fee. This would be useful for a disaster recovery site, 99.9% availability. After that we have one-zone IA, this option has 99.5% availability, is only in one AZ so if this fails the entire class fails, this option is useful if you need a copy of on-prem data that can be recovered. Now we go into the glacier classes, this is long-term storage. First off we have glacier instant retrieval, minimum storage time of 90 days and milliseconds to collect data. Follow that we have glacier flexible retrieval, also a minimum storage time of 90 days, however a retrieval time of up to 12 hours. Finally in the glacier section we have glacier deep archive, minimum of 180 days for storage time, up to 48 data retrieval. These glacier tiers get much cheaper deeper down. The most interesting is likely the intelligent tiering storage class, this automatically moves objects among tiers. First they start in standard, after 30 days they are moved into IA, after 90 days they are moved into glacier instant retrieval, optionally after 90 – 700+ they can be moved into glacier flexible retrieval, and finally 180 – 700+ they can move to glacier deep archive.