Not all files are used the same way: some you access daily, others almost never, but you still need to keep them. It would be absurd to pay the same for both. That's why S3 offers several storage classes: different "levels" with varying costs and access speeds. Choosing the right class can drastically reduce your storage bill.

The rule is simple:

The less you access a file, the cheaper it can be to store it… but retrieving it costs more time or money.

Analogy: Think about your belongings.

• What you use daily is on your desk (instant access, but desk space is "expensive"). → Standard.
• What you use occasionally you keep in a drawer (a bit more effort). → IA.
• What you almost never use but must keep (old documents) you put in boxes in the storage room (super cheap, but takes time to retrieve). → Glacier.

• For: frequently accessed data.
• Storage cost: the highest.
• Access: instant, no extra recovery cost.
• Typical use: files for an active website, app images, data that is read often.

• For: data you need to have instantly available, but access only a few times a month.
• Storage cost: cheaper than Standard.
• But: you pay a small fee each time you retrieve an object.
• Typical use: recent backups, files you might suddenly need but not daily.

• Same as Standard-IA but stores data in a single availability zone (instead of several).
• Cheaper, but if that AZ has a serious issue, you could lose the data.
• Typical use: data you can regenerate if lost (secondary backups, thumbnails you can recreate).

The Glacier family is for archiving data you almost never access but must keep (for example, for legal reasons). There are variants depending on how quickly you need to recover them:

• Typical use: historical files, backups from years ago, records that must be kept 7-10 years by law.

Real example: A hospital must keep medical records for 15 years by law, but rarely accesses them. They store them in Glacier Deep Archive: they pay a minimal amount to store them and, if they ever need one, accept waiting a few hours to recover it. The savings compared to Standard are huge.

If you don't know how often your data will be accessed, this class automatically moves each object to the cheapest tier based on usage. You do nothing; AWS optimizes the cost for you (in exchange for a small monitoring fee).

• Typical use: data with unpredictable or changing access patterns. It's a "safe and convenient" option when in doubt.

You can assign the class of an object when uploading it, or let them move automatically over time using lifecycle rules (we'll see this in subchapter 5.3). For example: "files move to IA after 30 days and to Glacier after 90". This way you optimize costs automatically without touching anything.

• S3 offers several storage classes that balance cost and speed/cost of access.
• Standard for day-to-day data; Standard-IA for infrequent access; Glacier for long-term archive (the deeper, the cheaper but slower to recover).
• One Zone-IA saves money at the cost of less resilience (regenerable data).
• Intelligent-Tiering lets AWS optimize cost automatically when you don't know the access pattern.
• Lifecycle rules move objects between classes automatically.

In the next subchapter we'll look at versioning (keeping multiple versions of an object) and lifecycle (automating class transitions and deletion), two key S3 features.