Glossary

Cloud storage

Cloud storage is storage delivered as a service over a network: requested through an API, drawn from a shared pool and charged by use. It describes how capacity is consumed, not where the hardware sits, and it can run in a public cloud, an owned data center or a sovereign cloud.

What makes storage cloud storage?

Four properties define it. Capacity is requested through an API, not a purchase order. It is drawn from a shared pool that spreads data across many machines. It can grow or shrink without a hardware project on the consumer's side. And use is measured, so charges follow what was stored and done. None of the four mentions a provider or a location, which is why the model can be delivered by a hyperscaler, by a regional provider or by an organization's own infrastructure.

Which types of cloud storage exist?

Three types come from the same pool. Object storage keeps data in buckets over HTTP and is addressed by name through an API. Block storage presents virtual disks to servers. File storage presents shared folders to many clients, as covered under network-attached storage. Object storage, through the S3 API, has become the common interface across clouds, which is why S3-compatible storage can be moved between environments with few application changes.

Why is placement the decision that matters?

The type follows from the workload. Where the capacity lives follows from how the data behaves, and three patterns recur as general guidance.

  • Demand that swings fits a public cloud, where capacity is paid for only while used and nothing is left idle.
  • Large, persistent data that is read often usually costs less on infrastructure an organization owns, because request and egress fees are charged on every read and grow with read volume while owned capacity is mostly a fixed cost.
  • Regulated data may need a sovereign cloud, where residency, jurisdiction and operator are set by law or contract.

The three are modeled side by side before commitment, using reads per month and the fees in egress fees as inputs. The same S3 bucket interface can sit in any of them, so placement can change without rewriting applications, a point explored under private cloud and hybrid cloud.

Frequently asked questions

Is cloud storage always in a public cloud?

No. The term describes the service model. The same API-driven, metered storage can run on infrastructure an organization owns or in a sovereign cloud.

What is the S3 API?

It is the request interface that originated with one provider's object storage and is now implemented by many systems. Applications written to it can usually switch between S3-compatible systems by changing an endpoint.

What drives the cost of cloud storage besides capacity?

Requests, retrievals from colder tiers, and data transferred out. These follow usage, so they vary month to month in a way capacity does not.