Amazon S3 Storage Classes Compared: Standard to Glacier Deep Archive

Amazon Simple Storage Service (S3) offers multiple storage classes that trade storage price against access cost and retrieval speed: S3 Standard for frequently accessed data, the Infrequent Access classes for data you touch rarely but need instantly, the Glacier classes for archives, and S3 Intelligent-Tiering for data whose access pattern you cannot predict. Every class except One Zone options stores objects redundantly across multiple Availability Zones with the same eleven nines of durability. Picking the right class is a cost decision, not a durability decision, and that is exactly how the CLF-C02 exam frames it.

What a storage class is

A storage class is a per-object attribute in S3. Objects in the same bucket can use different classes, and you can change an object’s class over time. Cheaper storage classes charge less per gigabyte stored but add retrieval fees, minimum storage durations, and in the Glacier tiers, retrieval delays. The mental model: the less often you read the data, the cheaper the class you should park it in.

The classes, from hot to cold

S3 Standard. The default. Low latency, high throughput, no retrieval fee, no minimum storage duration. Use it for active data: website assets, mobile app content, analytics working sets.

S3 Standard-Infrequent Access (Standard-IA). Lower storage price than Standard, same millisecond access, but you pay a per-gigabyte retrieval fee and a 30-day minimum storage duration. Meant for data accessed maybe once a month but needed immediately when requested: backups, older documents, disaster recovery files.

S3 One Zone-Infrequent Access (One Zone-IA). Like Standard-IA but stored in a single Availability Zone, which cuts the storage price further at the cost of resilience: if that Availability Zone is destroyed, the data is lost. Appropriate only for re-creatable data, such as secondary backup copies or thumbnails you can regenerate.

S3 Intelligent-Tiering. Automatically moves each object between access tiers based on observed access patterns, for a small per-object monitoring fee and no retrieval fees. Objects not accessed for 30 days move to an infrequent tier, then a deeper archive-instant tier after 90 days, and optional opt-in tiers go deeper still. This is the answer whenever access patterns are unknown, unpredictable, or changing.

S3 Glacier Instant Retrieval. Archive-priced storage that still returns data in milliseconds. Designed for long-lived data accessed about once a quarter, such as medical images or news archives. Minimum storage duration of 90 days.

S3 Glacier Flexible Retrieval. Cheaper still, but retrieval is asynchronous: expedited retrievals take minutes, standard retrievals take hours, and free bulk retrievals take longer. Suited to archives where a wait of minutes to hours is acceptable. 90-day minimum.

S3 Glacier Deep Archive. The lowest-cost storage in AWS. Retrievals take hours (standard) to half a day or more (bulk), with a 180-day minimum storage duration. Built for compliance archives and data retained for many years that you expect to read rarely, if ever.

S3 Express One Zone. The opposite end of the spectrum: a high-performance, single-Availability-Zone class delivering single-digit millisecond access for latency-critical workloads. It appears less often on CLF-C02 but is worth recognizing by name.

Class	Access speed	Minimum duration	Resilience	Typical use
Standard	Milliseconds	None	Multi-AZ	Active, frequently read data
Standard-IA	Milliseconds	30 days	Multi-AZ	Monthly-touched backups, DR files
One Zone-IA	Milliseconds	30 days	Single AZ	Re-creatable infrequent data
Intelligent-Tiering	Milliseconds (auto-tiers)	None for auto tiers	Multi-AZ	Unknown or changing access patterns
Glacier Instant Retrieval	Milliseconds	90 days	Multi-AZ	Quarterly-accessed archives
Glacier Flexible Retrieval	Minutes to hours	90 days	Multi-AZ	Archives tolerating retrieval waits
Glacier Deep Archive	Hours	180 days	Multi-AZ	Long-term compliance retention

Lifecycle policies tie it together

You rarely pick one class forever. S3 Lifecycle rules transition objects automatically as they age: for example, keep logs in Standard for 30 days, move them to Standard-IA, then to Glacier Flexible Retrieval after 90 days, and expire them after seven years. Lifecycle rules can also delete expired objects and clean up incomplete multipart uploads. On the exam, “automatically move older data to cheaper storage” maps to Lifecycle policies; “automatically adapt to unpredictable access” maps to Intelligent-Tiering. The distinction is age-based automation versus access-pattern-based automation.

Durability versus availability

All S3 classes are designed for eleven nines (99.999999999 percent) of object durability, meaning the chance of losing an object is vanishingly small. What varies between classes is availability (how often the data is reachable when requested), retrieval speed, and cost. One Zone classes are the exception on resilience: durability engineering is the same within the zone, but the data does not survive the loss of that Availability Zone. Exam questions sometimes bait you into thinking Glacier is “less durable”; it is not, it is just slower and cheaper to access.

S3 itself is object storage, one of three fundamentally different AWS storage models. If you are unsure when S3 is even the right service versus block or file storage, see EBS vs EFS vs S3. Storage class economics also feed the broader cost picture covered in AWS pricing models explained.

How the CLF-C02 exam tests this

• The retrieval-time discriminator. A scenario states data is rarely accessed but must be available “immediately” or “within milliseconds” when needed. That single word eliminates Glacier Flexible Retrieval and Deep Archive; the answer is Standard-IA or Glacier Instant Retrieval depending on how rare the access is.
• The lowest-cost archive pattern. Regulatory or compliance data must be kept for 7 to 10 years and retrieval within hours (or a day) is acceptable. The answer is S3 Glacier Deep Archive, the cheapest class.
• The unknown-access-pattern pattern. The question says access patterns are “unpredictable”, “unknown”, or “changing”, and asks how to optimize cost without operational effort. The answer is S3 Intelligent-Tiering, not a Lifecycle rule, because Lifecycle rules assume you already know the pattern.
• The re-creatable data pattern. A company wants the cheapest millisecond-access storage for data it could regenerate, such as processed image derivatives. The answer is One Zone-IA, and the justification is that single-AZ risk is acceptable for reproducible data.

Quick reference

• All classes target eleven nines of durability; class choice is about cost and access speed, not data loss risk.
• Standard: no retrieval fee, no minimum duration, hot data.
• Standard-IA: instant access, retrieval fee, 30-day minimum; One Zone-IA is the same in a single AZ for re-creatable data.
• Intelligent-Tiering: automatic tiering by access pattern, small monitoring fee, no retrieval fees; the “unpredictable access” answer.
• Glacier Instant Retrieval: archive price, millisecond access, 90-day minimum.
• Glacier Flexible Retrieval: minutes-to-hours retrieval, 90-day minimum.
• Glacier Deep Archive: cheapest, hours-long retrieval, 180-day minimum; the “long-term compliance” answer.
• Lifecycle rules automate age-based transitions between classes and object expiration.