Glossary

Object storage

Object storage is a way of storing data as individual objects, each containing the data itself, descriptive metadata and an identifier. Applications use that identifier to store or retrieve an object through an API. This makes object storage useful for growing collections of documents, images, video, backup data and AI datasets that applications need to access across a network.

For an IT team, the important distinction is how applications reach the data. An application requests an object by name, while the storage system manages where its contents live on the underlying hardware. That separation allows a distributed object storage platform to add capacity or move data between devices without requiring the application to track those physical changes.

How does object storage work?

Consider an inspection system that saves photographs from a manufacturing line. Each photograph can become an object, with metadata describing its content type and application-supplied details such as the inspection reference. The application records the object's identifier so it can retrieve the photograph when someone reviews the inspection. The storage platform handles placement and protection behind that request.

In an S3-based system, objects are organized into containers called buckets, and each object has a name called a key. A key might include slashes that make it look like a folder path, but those characters are generally part of the name. The application sends requests to write, read or delete the object through the S3 API. Access permissions determine which requests are allowed.

Metadata helps describe the object, but it does not automatically make every object searchable by its contents. Finding all photographs associated with a failed inspection may require an application database or a separate index. Object storage holds and returns the photographs; the inspection application supplies the business context that connects them.

How is object storage different from file and block storage?

File storage presents files and directories that applications can open through a file system. Block storage presents volumes that operating systems or databases use to read and write blocks of data. Object storage presents named objects accessed through an API. These interfaces support different application behaviors, even when they ultimately use similar drives underneath.

That difference matters when an application expects to update small portions of a file repeatedly or relies on shared-file locking. A conventional object API generally replaces an object's contents rather than providing the same in-place editing behavior as a file system. Some platforms offer additional capabilities, but an application built for a mounted file share does not automatically work with an object endpoint. Application support is therefore part of deciding whether object storage fits.

Why do organizations use object storage?

Object storage is useful when an organization needs to retain a growing body of data and make it available to multiple applications. Distributed platforms can spread that data across many storage nodes while presenting a common access service. Applications do not need a separate connection to each drive or server. This helps infrastructure teams expand storage without making the physical layout part of every application's configuration.

The inspection example illustrates the value. Photographs may accumulate for years, while quality teams, analytics applications and machine-learning pipelines need access to different subsets. Object storage can provide the shared repository beneath those uses. Permissions and application catalogs still govern who can retrieve the data and how they find the right objects.

Common uses include backup repositories, media libraries, research datasets and the storage layer of a data lake. These workloads often create data that will be read again later, rather than continuously edited in place. Object storage can also serve active analytics and AI workloads when its performance matches their access patterns. The storage model alone does not determine how fast those workloads will run.

How does object storage protect data?

Many object storage platforms distribute redundant data across devices or nodes so that a hardware failure does not destroy the only available copy. Some use replication, which keeps additional copies. Others use erasure coding, which stores data fragments and recovery information that can reconstruct missing fragments. The protection configuration determines which failures the system can tolerate and how much capacity protection consumes.

Hardware protection does not, by itself, prevent an authorized deletion. Features such as versioning and Object Lock address different risks: versioning can preserve earlier object versions, while retention controls can prevent protected versions from being deleted or overwritten during a defined period. These features need appropriate support and configuration. Storing data as objects does not automatically make it immutable or turn the platform into a complete backup strategy.

Does object storage have to run in the cloud?

Object storage can run in a public cloud, in an organization's own data center or in a service provider's environment. S3 is an access interface, so using it does not require storing data in Amazon's cloud. Applications can also connect to an on-premises platform that supports the S3 operations they need. Compatibility should be checked against the application's actual requirements.

Deployment location changes who operates the infrastructure and how data reaches applications. Keeping storage near the workloads that read it can reduce dependence on wide-area connections. Public cloud services offer a different operating model, with the provider managing the underlying infrastructure. The choice depends on access needs, operational resources and where the organization needs to keep its data.

Object storage with Scality RING

Scality RING provides distributed object storage for large enterprise, cloud and service-provider environments. It supports S3 access and distributes storage across infrastructure managed by the operator. Organizations can use it to provide a shared storage service for growing data collections while controlling the deployment. The application-facing object interface remains separate from the hardware used to hold and protect the data.