Object storage is a storage infrastructure with which unstructured data is managed in the form of objects that are not organized hierarchically in folders. In contrast, files are managed hierarchically in file storage, and in block storage, data is stored in data blocks on tracks and sectors.
In object storage, each individual object is stored as a binary large object( BLOB), and is also uniquely described by metadata and identified by a unique global identifier. As a result, object stores do not require a hierarchical structure. The namespace of objects has a simple structure and consists of one level. Objects cannot be placed within other objects. The dependency between objects is eliminated and the basic functions of object storage, namely storing and retrieving objects, are improved.
Object stores have enormous storage capacities and are predestined for Big Data with data volumes of petabytes and exabytes. Powerful search algorithms support easy and efficient storage management and provide much faster access compared to other storage architectures. Object storage supports capabilities that cannot be realized with any other storage architecture. For example, interfaces that are programmed directly by the application or a namespace that can be spanned across many hardware instances and management functions such as data replication or data distribution.
Object storage is supported by various the S3 interface from Amazon and storage protocols such as the OpenStack Foundation's Swift protocol. Object storage can be used as a cloud service, or as a converged storage array with storage network or NAS storage.