Think of it this way: Block is the first, most basic and most primitive of all the data storage structures. It is used to define the building blocks of storage (e.g. sectors) of any single hard drive or group of hard drives via SATA, SCSI, SAS or USB on any local system. Where available on a network or LAN, block storage would be defined over protocols such as ATA over Ethernet, iSCSI, or Fibre Channel to target large groups of disks collectively known as a Storage Area Network (SAN). When a disk or group of disks are formatted or initialized the block structure is determined and written to the targeted storage.

File storage is next. This data storage type is most familiar to end users. A formatted disk or disks contains a file system which is usually but not always a hierarchical based file system which sets up resources to store logical entities known as files, directories, sub-directories and folders. Sharing data files at this level can be performed by computer networking protocols: SMB, CIFS, NFS and AppleTalk. The resulting server is often called a Network-Attached Storage (NAS) device.

Object storage is the new kid on the playground. It builds upon block and file methods but it's not necessarily limited to the disadvantages of either. What we are describing here is a higher level of data abstraction which gives the user the ability to perceive and use data without coupling it with direct, physical hardware assets. Object storage is optimized for the web and cloud, can be easily clustered, scales up to very high data capacities and can scale across geographies (given certain limits). With S3 object storage, data is logically grouped into buckets which would contain the objects (files). Instead of sub-directories, prefixes can be used for deeper levels of separating or subdividing content.