Object Storage: The Cloud's Workhorse
Of the three storage types, one shows up almost everywhere: object storage. It's the cheap, near-bottomless pool where the cloud keeps photos, videos, backups, and most of what a website serves to its visitors.
Unlike a hard drive or a shared folder, object storage doesn't look like a file system. Think of a vast valet coat-check: you hand over a labeled item, they store it safely across multiple locations, and you retrieve it by the label from anywhere. You never touch the shelves yourself — you just use the claim ticket. That claim ticket is the object's name, and the coat-check is the bucket.
Buckets and Objects
Object storage organizes files into buckets — named containers you create, a bit like top-level labeled bins. Inside a bucket, each file is called an object: a photo, a video, a log file, a compressed backup, or any other file. You name each object when you upload it and retrieve it later by that name.
There is no subfolder hierarchy in the traditional sense. Names can look like paths ("images/2024/logo.png") but the storage service treats the whole thing as one flat name. The structure is a convenience, not a real directory tree.
Effectively Unlimited and Cheap
Unlike a physical disk you buy and can fill, a bucket holds as much as you put in it. You pay for the gigabytes you actually store — not for reserved capacity you may never fill. A startup with a dozen files and a streaming service with billions of videos use the same bucket model. The storage grows automatically; you never buy a bigger disk.
Durable by Design
Cloud providers keep multiple copies of every object across separate, physically distinct buildings. If one building loses power or suffers a hardware failure, your file still exists, and the provider serves a copy from another location transparently.
This is often expressed as "eleven nines" of durability — 99.999999999%. In plain terms: across a year, the chance of losing a single file is extraordinarily small. That figure is a simplification; the full reliability picture also involves how you use backups and versioning. But the point holds: object storage is designed from the ground up to not lose your files.
Reached Over the Web
Objects are reached through the storage service's API, usually over HTTPS — and access is controlled by permissions. Objects can be public or private, and are private by default; a URL alone does not mean public access. When an object is explicitly made public, it gets its own web address that any browser can load directly, without a separate web server in the way. This makes object storage the natural home for anything a website or app needs to deliver to users: product images, downloadable files, videos, and the HTML of static websites themselves.
- "A bucket is just a folder on a server." A bucket is a separate web service, not part of any server's file system. Objects are reached by name over the internet — you don't mount a bucket the way you mount a drive or a folder.
- "Object storage has a size limit I'll eventually hit." Buckets grow without a fixed capacity ceiling. You pay per gigabyte used; there is no disk to fill up.
- "Object storage is a kind of database." It stores whole files by name — not rows or queryable fields. You can't filter "show me all photos tagged 'sunset'" the way a database handles queries. For structured data you query, you need a database; for whole files you fetch by name, you use object storage.
- "Eleven nines means object storage never loses data." It means loss is extraordinarily unlikely, not mathematically impossible. Versioning and cross-region replication add additional safety on top, and are worth using for critical data.
- Object storage is behind most of what you see on the web — images, downloadable files, streamed video, and static sites. Understanding it means understanding the storage layer of the modern internet.
- Backups and data archives almost always land in object storage first; it's the default landing zone for data before it's analyzed, moved, or deleted.
- Knowing S3, Cloud Storage, and Blob Storage as names for the same bucket-and-object model means you can follow any cloud architecture discussion regardless of which provider is being used.
Knowledge Check
What is a "bucket" in object storage?
- A named container that holds objects (files)
- A virtual disk attached to one server at a time
- A shared network folder many machines can mount simultaneously
- A scheduled process that moves files between storage tiers
Why is object storage described as "effectively unlimited"?
- You pay per gigabyte stored; there is no fixed capacity to fill
- It provides the fastest read speeds of any storage type available
- It only stores files smaller than a defined size limit per object
- Objects are automatically deleted after 30 days to free up space
What does "eleven nines of durability" mean for stored objects?
- Files are copied across separate buildings, making loss very unlikely
- Objects expire automatically and must be re-uploaded every eleven years
- The bucket can hold exactly eleven billion files before it fills up
- Each file is guaranteed to be retrieved in under eleven seconds
You got correct