CLF-C02 · SAA-C03
S3 storage classes explained
What each Amazon S3 storage class is for, how retrieval and minimum-duration charges work, and how to pick one on the CLF-C02 and SAA-C03 exams.
Why S3 has storage classes
Every object you put in Amazon S3 has a storage class. The class decides how much you pay to keep the object, how much you pay to read it back, how fast you can read it, and how many Availability Zones hold copies of it.
The data and the API stay the same. A PDF in S3 Standard and the same PDF in S3 Glacier Deep Archive are both objects in a bucket. What changes is the trade between storage cost and access cost. Cheaper storage almost always means slower or more expensive retrieval.
Exam questions on this topic nearly always describe an access pattern and ask for the cheapest class that still meets it. So the skill to learn is reading the pattern.
The classes you need to know
Here is each class with the access pattern it fits. All of them except the One Zone classes store data across at least three Availability Zones.
- S3 Standard: frequently accessed data, no retrieval fee, millisecond access. The default when you upload.
- S3 Intelligent-Tiering: unknown or changing access patterns. S3 moves each object between access tiers based on how often it is read, for a small per-object monitoring fee and no retrieval fee.
- S3 Standard-Infrequent Access (Standard-IA): data read rarely but needed in milliseconds when it is, such as backups you might restore. Lower storage price, per-GB retrieval fee.
- S3 One Zone-Infrequent Access: like Standard-IA but kept in a single Availability Zone. Cheaper, and lost if that zone is destroyed. Good for data you can recreate, such as thumbnails.
- S3 Express One Zone: the fastest class, single-digit millisecond latency in one Availability Zone you choose, stored in directory buckets. Used for ML training data and other request-heavy workloads.
- S3 Glacier Instant Retrieval: archive data read about once a quarter that still needs millisecond access, such as old medical images.
- S3 Glacier Flexible Retrieval: archives where waiting minutes to hours is fine. Offers expedited, standard and free bulk retrievals.
- S3 Glacier Deep Archive: the lowest storage price, with retrieval measured in hours. Built for compliance records kept for years and almost never read.
Costs that catch people out
The infrequent access and Glacier classes have a minimum storage duration. If you delete or move an object before that period ends, you are billed for the rest of it anyway. Standard-IA, One Zone-IA and Glacier Instant Retrieval also have a minimum billable object size, so millions of tiny files can cost more in Standard-IA than in Standard.
Retrieval fees matter too. Standard-IA looks cheaper on paper, but if you read the data every day, the per-GB retrieval charges can make it more expensive than S3 Standard. That is why the exam keeps pairing "infrequently accessed" with these classes and "frequently accessed" with Standard.
Intelligent-Tiering avoids retrieval fees entirely. Very small objects are not monitored and always stay in the Frequent Access tier, so it suits larger objects best.
Archived objects are not instantly readable
Objects in Glacier Flexible Retrieval and Glacier Deep Archive cannot be read with a normal GET request. You first start a restore, wait for it to finish, and then read a temporary copy. Glacier Instant Retrieval is the exception: it reads like any other S3 object.
If a question says users must be able to download an archived file "immediately" or "in milliseconds", Flexible Retrieval and Deep Archive are wrong answers, however cheap they are.
Moving data between classes with lifecycle rules
You rarely pick one class forever. S3 Lifecycle rules move objects to cheaper classes as they age and can delete them at the end. A typical rule for application logs might look like this:
- Day 0: upload to S3 Standard while the team is still debugging.
- After 30 days: transition to Standard-IA.
- After 180 days: transition to Glacier Deep Archive for compliance.
- After 7 years: expire (delete) the objects.
How to answer exam questions
Look for three clues: how often the data is read, how fast it must come back, and whether it can be recreated. Frequent access points to Standard. Unknown or shifting access points to Intelligent-Tiering. Rare access with instant reads points to Standard-IA or Glacier Instant Retrieval. Long-term archives with hours of tolerance point to Glacier Flexible Retrieval or Deep Archive. Recreatable data that can live in one zone points to One Zone-IA.
Durability is not a tiebreaker. S3 classes are all designed for the same very high object durability; One Zone classes just store that data in a single zone, so they do not survive the loss of that zone. Availability, not durability, is what differs most between classes.
Practice questions
A financial firm must keep records for several years for compliance. The records will almost never be read, and a retrieval time of up to 48 hours is acceptable. Which S3 storage class has the lowest storage cost for this need?
- S3 Glacier Instant Retrieval
- S3 Glacier Deep Archive
- S3 Standard-Infrequent Access
- S3 Intelligent-Tiering
Answer
B. S3 Glacier Deep Archive is the lowest-cost S3 storage class, intended for long-term retention where retrieval within hours is acceptable. Glacier Instant Retrieval and Standard-IA provide millisecond access at a higher storage price, and Intelligent-Tiering is for unpredictable access patterns.
CLF-C02 question 265
A company stores data in Amazon S3 but has no idea how often each object will be accessed, and access patterns change over time. Which storage class automatically moves objects between access tiers to save cost?
- S3 Intelligent-Tiering
- S3 Glacier Deep Archive
- S3 Standard
- S3 One Zone-Infrequent Access
Answer
A. S3 Intelligent-Tiering monitors access and automatically moves objects between tiers to lower cost when access patterns are unknown or changing. S3 Standard keeps everything in one frequently accessed tier, while One Zone-IA and Glacier Deep Archive suit data whose infrequent access is already known.
CLF-C02 question 162
A research institute archives sensor readings to Amazon S3 Glacier Deep Archive. Each reading is uploaded as its own object of a few kilobytes, and there are several hundred million objects. The data is almost never restored. The bill is much higher than the institute expected from the total volume of data. Which change will reduce the storage cost the MOST?
- Enable S3 Versioning on the bucket that holds the readings.
- Store each reading as its own object in S3 Glacier Instant Retrieval instead.
- Combine the small readings into large archive files before uploading them to S3 Glacier Deep Archive.
- Store each reading as its own object in S3 Standard-Infrequent Access instead.
Answer
C. Every object in the Glacier Flexible Retrieval and Deep Archive storage classes carries a fixed amount of extra billed metadata, and requests are charged per object, so hundreds of millions of tiny objects cost far more than the same data stored as a small number of large archives. Glacier Instant Retrieval and Standard-IA bill small objects at a minimum object size, so they do not solve the problem.
SAA-C03 question 403
A company writes application logs to Amazon S3. Logs are queried heavily for the first 30 days, must be readable immediately but are rarely read for the rest of the first year, and must then be kept for 7 years in total for audits. Auditors give at least 3 days of notice before they need old logs. Which lifecycle configuration is MOST cost-effective?
- Keep logs in S3 Standard, transition them to S3 Standard-IA after 30 days, transition them to S3 Glacier Deep Archive after 365 days, and expire them after 7 years.
- Keep logs in S3 Standard, transition them to S3 Glacier Deep Archive after 30 days, and expire them after 7 years.
- Keep logs in S3 Standard, transition them to S3 Glacier Instant Retrieval after 30 days, and keep them there until they expire after 7 years.
- Keep logs in S3 Standard for the full 7 years and expire them after 7 years.
Answer
A. Standard-IA keeps logs immediately readable at a lower storage price during the rarely accessed first year, and Glacier Deep Archive is the cheapest class for the long audit period because a 3-day notice easily covers its retrieval time. Moving to Deep Archive after only 30 days breaks the requirement for immediate reads during the first year. Keeping six years of audit-only data in Glacier Instant Retrieval works but costs much more than Deep Archive.
SAA-C03 question 32