S3 Storage Classes

Overview

S3 offers 7 storage classes designed for different use cases. They differ in cost, availability, retrieval time, and minimum storage duration. The key insight: you're trading access speed and availability for lower cost. The less frequently you access data, the cheaper it is to store.

You can choose the storage class when uploading an object, change it manually later, or use Lifecycle Rules to automate transitions between classes.

Durability vs Availability — Know the Difference

These two concepts are fundamental and frequently tested:

Durability (How Safe Is My Data?)

Durability measures the probability of NOT losing your data. S3 achieves this by automatically replicating your data across multiple devices and facilities within a region.

Availability (Can I Access My Data Right Now?)

Availability measures how readily the service is accessible. Lower availability means you might occasionally get errors when trying to access your data, and your application needs to handle that.

The 7 Storage Classes in Detail

1. S3 Standard (General Purpose)

The default storage class. Use this when you don't know what to pick or when you need frequent access.

Use cases: Big data analytics, mobile and gaming applications, content distribution, frequently accessed data.

2. S3 Standard-Infrequent Access (Standard-IA)

For data that you don't access often, but when you do, you need it immediately (low latency). Cheaper storage cost than Standard, but you pay a retrieval fee per GB retrieved.

Use cases: Disaster recovery files, backups that you rarely need but must access quickly when you do.

3. S3 One Zone-Infrequent Access (One Zone-IA)

Same as Standard-IA but stored in only one Availability Zone. This makes it cheaper, but if that AZ is destroyed (fire, flood, earthquake), your data is lost.

Use cases: Secondary copies of backups (primary backup is elsewhere), data you can recreate if lost (e.g., thumbnails generated from original images).

4. S3 Glacier Instant Retrieval

Archive storage with millisecond retrieval. Think of it as "cold storage that you can access instantly." Great for data accessed about once per quarter.

Use cases: Medical images accessed quarterly, financial reports accessed once per quarter, compliance data that must be retrievable instantly.

5. S3 Glacier Flexible Retrieval

Archive storage where you're willing to wait minutes to hours to get your data back. This used to be called just "Amazon S3 Glacier" before they added more tiers.

Use cases: Backup data accessed semi-annually, disaster recovery data where you can wait a few hours, long-term data retention.

6. S3 Glacier Deep Archive

The cheapest storage class in S3. Designed for data you almost never access — think years of archived records. You must be willing to wait 12 to 48 hours to retrieve your data.

Use cases: Regulatory archives (7+ years), compliance data, data that's legally required to be kept but almost never accessed.

7. S3 Intelligent-Tiering

The "set it and forget it" storage class. S3 automatically moves your objects between access tiers based on how frequently you access them. You pay a small monthly monitoring fee, but there are no retrieval charges.

Intelligent-Tiering Access Tiers

Object uploaded
       │
       ▼
┌─────────────────────────────┐
│ Frequent Access (default)   │ ← automatic
└──────────────┬──────────────┘
               │ Not accessed for 30 days
               ▼
┌─────────────────────────────┐
│ Infrequent Access           │ ← automatic
└──────────────┬──────────────┘
               │ Not accessed for 90 days
               ▼
┌─────────────────────────────┐
│ Archive Instant Access      │ ← automatic
└──────────────┬──────────────┘
               │ Configurable: 90–700+ days
               ▼
┌─────────────────────────────┐
│ Archive Access (optional)   │ ← must opt-in
└──────────────┬──────────────┘
               │ Configurable: 180–700+ days
               ▼
┌─────────────────────────────┐
│ Deep Archive Access (opt.)  │ ← must opt-in
└─────────────────────────────┘

The first three tiers are automatic. The last two (Archive Access and Deep Archive Access) are optional and must be configured.

Use cases: When you don't know your access patterns or when access patterns change over time. Let AWS optimize the cost for you.

Storage Class Decision Tree

How often do you access the data?
│
├── Frequently (daily/weekly)?
│   └── S3 Standard
│
├── Infrequently but need instant access when you do?
│   ├── Need multi-AZ durability?
│   │   └── S3 Standard-IA
│   └── Single AZ OK? (data is recreatable)
│       └── S3 One Zone-IA
│
├── Rarely (archive)?
│   ├── Need millisecond retrieval?
│   │   └── Glacier Instant Retrieval (90-day min)
│   ├── Can wait 1 min – 12 hours?
│   │   └── Glacier Flexible Retrieval (90-day min)
│   └── Can wait 12 – 48 hours?
│       └── Glacier Deep Archive (180-day min)
│
└── Don't know / patterns change?
    └── S3 Intelligent-Tiering

Minimum Storage Duration — Important for Cost

If you delete, overwrite, or transition an object before the minimum storage duration, you're still charged for the full duration.

Lifecycle Rules — Automating Transitions

Instead of manually changing storage classes, you can create Lifecycle Rules that automatically transition objects based on age:

Example Lifecycle Rule:
┌──────────────┐  30 days  ┌──────────────┐  60 days  ┌───────────────────┐
│ S3 Standard  │─────────▶│ Standard-IA  │─────────▶│ Intelligent-      │
└──────────────┘          └──────────────┘          │ Tiering           │
                                                     └────────┬──────────┘
                                                              │ 180 days
                                                              ▼
                                                     ┌───────────────────┐
                                                     │ Glacier Flexible  │
                                                     │ Retrieval         │
                                                     └────────┬──────────┘
                                                              │ 365 days
                                                              ▼
                                                     ┌───────────────────┐
                                                     │ DELETE (expire)   │
                                                     └───────────────────┘

Creating a Lifecycle Rule (Hands-On)

1. Go to bucket → Management tab → Lifecycle rules → Create rule
2. Name: DemoRule
3. Scope: apply to all objects in the bucket (or filter by prefix/tags)
4. Choose Move current versions between storage classes
5. Set transitions:
   • Standard-IA after 30 days
   • Intelligent-Tiering after 60 days
   • Glacier Flexible Retrieval after 180 days
6. Optionally: Expire current versions after 365 days (auto-delete)
7. Review the transition timeline → Create rule

Hands-On: Changing Storage Classes

At Upload Time

1. Upload a file → expand Properties section
2. Under Storage class, you'll see all available classes with their details
3. Choose the desired class (e.g., Standard-IA) → Upload

After Upload

1. Click on the object → Properties tab → scroll to Storage class
2. Click Edit → choose a new class (e.g., One Zone-IA, Glacier Instant Retrieval)
3. Save changes

Questions & Answers