Glossary›Storage Systems, Protocols, and Scale›Cloud storage

Cloud storage

Why does cloud storage matter for unstructured data?

Because more and more file data lives in the cloud or passes through it: backups, archives, media libraries, data lakes, and AI pipelines. Cloud storage lets an organization add capacity in minutes instead of buying hardware, which matters most when hardware prices and lead times rise. It also spreads data across more places, so knowing what sits where becomes harder.

What is the difference between cloud storage and on-premises storage?

On-premises storage is hardware an organization buys, runs, and keeps in its own data center. Cloud storage is rented from a provider and paid for by use. The main differences are cost, since the cloud charges for capacity, requests, and reading data out, while on-premises storage costs more up front; speed, since local storage sits closer to the users and applications that read it; and control, since the provider runs the hardware and the data sits in the provider’s region. Most organizations use both, which is called hybrid cloud storage.

What is an example of cloud storage in use?

A studio keeps active projects on its own fast file storage. When a show is delivered, its files move to cloud object storage, and after a year with no access, to a cloud archive tier that costs very little to keep. Editors can still search for any file across all three places, and bring one back when a sequel starts.

What kinds of cloud storage are there?

Providers offer four main kinds:

• Object storage, such as Amazon S3, Azure Blob Storage, and Google Cloud Storage, which holds most cloud data.
• File storage, which offers managed network shares that applications open by path.
• Block storage, which gives virtual servers their disks.
• Archive tiers, which cost very little to keep data in but charge more, and take longer, to read it back.

What does cloud storage really cost?

The monthly price per terabyte is only part of the bill. Providers also charge for requests, for reading data back out, called egress, and for retrieving data from archive tiers, and some tiers bill a minimum storage period even if data is deleted early. Wasabi’s 2026 Global Cloud Storage Index found that 50% of cloud storage spending goes to fees, and that 49% of organizations went over their cloud storage budget in 2025. Knowing what you hold, and how often it is read, before you move it is what keeps those fees down.

How do you see what you hold across several clouds and on-premises systems?

By indexing the metadata of every bucket, container, and file system into one catalog. Each cloud provider exposes object metadata through its API, so a scanner can list objects with their size, dates, and storage class, next to the files on premises. One view shows duplicate copies across clouds, buckets nobody owns, and data sitting on an expensive tier that is never read.

Are there limits to indexing cloud storage compared with on-premises storage?

Yes, a few, because cloud object storage works differently from a file system:

• Fewer fields: object storage records size, last modified date, storage class, and tags, but usually no owner and no last access date, so “cold” has to be judged another way.
• Listing costs money: providers charge for each listing request, so scanning billions of objects has a small but real cost.
• Rate limits: providers slow down clients that send too many requests at once, which sets a ceiling on scan speed.
• No folders to compare: without folder modification dates, a rescan cannot skip unchanged areas as easily, so providers’ inventory reports or change notifications are often used to keep the index current.
• Archive tiers: objects in deep archive can be listed, but their content cannot be read until they are restored.

None of these stops cloud storage from being indexed, but they shape how often and how deeply it is scanned.

How do you move data to the cloud and back?

Choose what to move from metadata first, so only data that belongs in the cloud goes there. Then copy it with a tool that verifies each file arrived intact, and keep the catalog current so users can still find it. Moving data back from the cloud, called repatriation, follows the same steps, but budget for egress fees, which can be large for big data sets.

Where is cloud data physically stored?

In the provider’s data centers, in the region chosen when each bucket or share was created, such as a region in the United States or the European Union. The region is recorded in the storage settings, so a catalog that records it for every object can show which data sits in which country. That matters for data sovereignty rules, which can require personal or regulated data to stay in a specific country.

← Back to Storage Systems, Protocols, and Scale

Last updated: November 13, 2026.