This chapter dives deep into S3 Intelligent-Tiering, one of the most powerful cost-optimization features for Amazon S3. For the SAA-C03 exam, questions on Intelligent-Tiering appear in roughly 5–10% of tests, usually as part of a larger scenario about optimizing storage costs for unpredictable or unknown access patterns. Understanding exactly how the four access tiers work, the monitoring fee structure, and the 30-day auto-tiering delay is critical to selecting the right storage class for a given workload and avoiding costly mistakes.
Jump to a section
Imagine you live in a house with four distinct climate zones: a rarely-used guest bedroom (accessed maybe once a year), a home office used daily but only during business hours, a living room used frequently but not every day, and a kitchen used multiple times daily. You install a smart thermostat that continuously monitors motion sensors in each room. When a room is not used for a set period, the thermostat automatically adjusts the temperature to save energy—turning off the AC entirely for the guest bedroom, reducing heating in the living room after 30 days of no motion, and keeping the office at a comfortable level only during work hours. The thermostat never asks you to set rules; it learns patterns and adapts. You pay a small monthly fee for the smart thermostat, but your energy bill drops significantly because you no longer heat or cool unused space at full cost. If someone unexpectedly enters the guest bedroom, the thermostat quickly restores comfortable temperature within minutes. This is exactly how S3 Intelligent-Tiering works: it monitors access patterns of objects, automatically moves them between four access tiers (Frequent, Infrequent, Archive Instant, and Deep Archive) based on how often they are accessed, charges a small monitoring fee per object, and ensures that if an object is accessed after being moved to a colder tier, it is instantly available (with no retrieval delay for the first three tiers). You never manually define lifecycle rules—the service optimizes cost automatically.
What is S3 Intelligent-Tiering and Why Does It Exist?
Amazon S3 Intelligent-Tiering is an S3 storage class designed for data with unknown, changing, or unpredictable access patterns. It automatically moves objects between four access tiers based on changing access patterns, optimizing storage costs without performance impact or operational overhead. The four tiers are:
Frequent Access Tier (default upon upload)
Infrequent Access Tier (after 30 days of no access)
Archive Instant Access Tier (after 90 days of no access)
Deep Archive Access Tier (after 180 days of no access)
The service charges a small monthly monitoring and automation fee per object (currently $0.0025 per 1,000 objects), but no retrieval fees for accessing objects in the Frequent, Infrequent, or Archive Instant tiers. Only the Deep Archive tier incurs a retrieval cost (standard Deep Archive retrieval fees apply).
How It Works Internally
When you upload an object using the S3 Intelligent-Tiering storage class, it is initially placed in the Frequent Access tier. S3 then monitors the access patterns of each object—specifically, the last access time. If an object is not accessed for 30 consecutive days, S3 automatically moves it to the Infrequent Access tier. If another 60 days pass without access (90 consecutive days total), it moves to the Archive Instant Access tier. After another 90 days without access (180 consecutive days total), it moves to the Deep Archive Access tier.
Importantly, if an object is accessed at any point, the clock resets: the object is promoted back to the Frequent Access tier, and the 30-day countdown restarts. This promotion is immediate and does not incur retrieval costs (except for Deep Archive, which requires a standard restore process). The entire process is transparent—you simply read the object as normal using GetObject, and S3 handles the tier transitions behind the scenes.
Key Components, Values, Defaults, and Timers
Monitoring fee: $0.0025 per 1,000 objects per month (approx $2.50 per million objects per month). This is charged regardless of whether the object is accessed or not.
Tier transition timers:
- Frequent -> Infrequent: 30 days of no access - Infrequent -> Archive Instant: 90 days of no access (cumulative from last access) - Archive Instant -> Deep Archive: 180 days of no access (cumulative from last access) - Retrieval costs:
- Frequent, Infrequent, Archive Instant: No retrieval fee - Deep Archive: Standard Deep Archive retrieval fees apply (per GB retrieved) - Minimum object size: 128 KB for the monitoring fee to be cost-effective. Objects smaller than 128 KB can be stored but may not save money due to the fixed monitoring fee. - Availability: Same as S3 Standard for Frequent tier; Infrequent and Archive Instant have slightly lower availability (99.9% vs 99.99%) but are still designed for 11 9s durability. - Transition time: Tier movements are not instant—they typically occur within 24-48 hours after the threshold is met.
Configuration and Verification
To use S3 Intelligent-Tiering, you simply specify the storage class when uploading an object:
aws s3 cp myfile.txt s3://my-bucket/ --storage-class INTELLIGENT_TIERINGYou can also set it as the default storage class for a bucket using lifecycle policies, but the recommended approach is to use it directly per object.
To verify the current tier of an object, use the head-object command:
aws s3api head-object --bucket my-bucket --key myfile.txtLook for the StorageClass field in the output, which will show INTELLIGENT_TIERING (the object's storage class) and the TieringSummary field (if enabled) showing the current access tier.
To check the access tier, enable S3 Inventory reports that include the IntelligentTieringAccessTier field, or use S3 Storage Lens.
Interaction with Related Technologies
S3 Lifecycle Policies: You can combine Intelligent-Tiering with lifecycle policies to eventually expire objects (delete them) after a certain period. For example, store in Intelligent-Tiering for 365 days, then expire.
S3 Object Lock: Intelligent-Tiering is compatible with Object Lock for write-once-read-many (WORM) compliance.
S3 Replication: Objects replicated to another bucket can also be stored with Intelligent-Tiering, but the replication destination bucket must also support the storage class.
S3 Glacier: Intelligent-Tiering includes Archive Instant and Deep Archive tiers, which are distinct from S3 Glacier Flexible Retrieval and S3 Glacier Deep Archive storage classes. The key difference is that Intelligent-Tiering manages transitions automatically, whereas Glacier classes are manual.
Limitations and Considerations
Not suitable for data with predictable access patterns: If you know your data is accessed infrequently from the start, use S3 Standard-IA or One Zone-IA directly to avoid the monitoring fee.
Minimum storage duration: There is no minimum storage duration charge for Intelligent-Tiering (unlike S3 Standard-IA which has a 30-day minimum). You pay only for what you use.
Object size: For objects smaller than 128 KB, the monitoring fee may outweigh savings; consider using S3 Standard for small objects.
No retrieval delay: Except for Deep Archive, objects are immediately available when accessed. Deep Archive requires a restore process that can take up to 12 hours for standard retrieval.
Monitoring fee per object: The fee applies to every object in the Intelligent-Tiering storage class, regardless of size. For large numbers of small objects, this can add up.
Upload Object to Intelligent-Tiering
When you upload an object using the S3 Intelligent-Tiering storage class, S3 places it in the Frequent Access tier by default. The object is stored redundantly across multiple Availability Zones (same as S3 Standard). S3 begins tracking the last access time for this object. A monitoring fee of $0.0025 per 1,000 objects per month starts accruing immediately. There is no minimum storage duration charge, so you can delete the object at any time without penalty.
30 Days of No Access Triggers First Move
If the object is not accessed (via GetObject, HeadObject, etc.) for 30 consecutive days, S3 automatically moves it from the Frequent Access tier to the Infrequent Access tier. This move is transparent—you do not need to update any pointers or metadata. The object remains immediately accessible with the same low latency. The storage cost drops to the Infrequent Access tier rate (approx $0.0125/GB vs $0.023/GB for Frequent). The monitoring fee continues.
90 Days of No Access Triggers Second Move
If the object continues to remain unaccessed for another 60 days (total 90 days from last access), S3 moves it to the Archive Instant Access tier. This tier offers even lower storage cost (approx $0.004/GB) but still provides immediate access with no retrieval delay. The object is still stored redundantly across AZs. The monitoring fee continues.
180 Days of No Access Triggers Final Move
After a total of 180 days without access (90 more days since the last move), S3 moves the object to the Deep Archive Access tier. This is the cheapest storage tier (approx $0.00099/GB) but with a trade-off: the object is no longer immediately accessible. To read the object, you must initiate a restore process (using RestoreObject API) that takes 12–48 hours for standard retrieval. Retrieval costs apply (per GB). The monitoring fee continues.
Access Resets the Clock and Promotes Tier
If at any point the object is accessed (GetObject, HeadObject, or any operation that reads the object), the clock resets to zero and the object is immediately promoted back to the Frequent Access tier. This promotion is automatic and incurs no retrieval fee (except if the object was in Deep Archive, in which case you must first restore it, incurring retrieval costs). After promotion, the 30-day countdown restarts. The monitoring fee continues throughout.
Scenario 1: Media Archive for a News Agency
A global news agency stores millions of video clips and images from past events. Access patterns are unpredictable—a clip might go viral months after publication, then lie dormant for years. Initially, they used S3 Standard, but costs ballooned to $50,000/month. By migrating to S3 Intelligent-Tiering, they reduced storage costs by 60% without any manual intervention. They configured a lifecycle policy to expire objects after 10 years. Performance remained excellent because frequently accessed clips (breaking news) stayed in the Frequent tier. The monitoring fee for 10 million objects was only $25/month. One mistake: they initially did not enable S3 Inventory to track which tier objects were in, making it hard to audit costs. They later enabled S3 Storage Lens to get visibility.
Scenario 2: User-Generated Content Platform
A social media platform stores user photos and videos. Most content is never accessed after the first week, but a small fraction becomes popular again later. They use Intelligent-Tiering to automatically move old content to cheaper tiers. However, they noticed that objects smaller than 128 KB (profile thumbnails) were costing more in monitoring fees than they saved. They switched thumbnails to S3 Standard and kept only larger files in Intelligent-Tiering. They also set up S3 Event Notifications to alert when objects move to Deep Archive, because restoring those for occasional access was too slow. They implemented a separate cache layer (CloudFront) for recently accessed Deep Archive objects.
Scenario 3: Healthcare Data Compliance
A healthcare company stores patient records that must be retained for 7 years. Access patterns are unpredictable—some records are accessed frequently during treatment, then never again. They use S3 Intelligent-Tiering with Object Lock in compliance mode to prevent deletion. They configured a lifecycle policy to transition objects to S3 Glacier Deep Archive (not the Intelligent-Tiering Deep Archive tier) after 2 years because they wanted to use the bulk retrieval option for lower cost. However, this required manual lifecycle rules. They later realized that using Intelligent-Tiering's Deep Archive tier would have been simpler, but the bulk retrieval option is not available for Intelligent-Tiering's Deep Archive tier (only standard and expedited). They had to adjust their strategy.
SAA-C03 Exam Focus on S3 Intelligent-Tiering
The SAA-C03 exam tests your ability to choose the right storage class for a given scenario. For Intelligent-Tiering, the exam focuses on: - Objective 4.2: Cost optimization strategies – selecting storage classes that minimize cost based on access patterns. - Objective 2.3: Data lifecycle management – understanding automatic tiering.
Common Wrong Answers and Why Candidates Choose Them
Choosing S3 Standard-IA for unknown patterns: Many candidates think Standard-IA is cheaper because it has no monitoring fee. However, Standard-IA has a 30-day minimum charge and retrieval fees, making it more expensive than Intelligent-Tiering if data is accessed frequently. Intelligent-Tiering is better for unknown patterns.
Assuming Intelligent-Tiering has retrieval fees for all tiers: Candidates often think moving to Infrequent tier incurs retrieval costs. In reality, only Deep Archive tier has retrieval fees; the other three tiers have no retrieval fees.
Believing Intelligent-Tiering is suitable for small objects: Candidates may not consider the monitoring fee per object. For objects under 128 KB, the monitoring fee can exceed savings. The exam may present a scenario with many small objects and ask for the most cost-effective solution.
Confusing Intelligent-Tiering with S3 Lifecycle policies: The exam tests that Intelligent-Tiering is a storage class, not a lifecycle policy. Lifecycle policies are rules you define; Intelligent-Tiering is automatic.
Specific Numbers and Terms to Memorize
Monitoring fee: $0.0025 per 1,000 objects per month
Tier transition thresholds: 30 days, 90 days, 180 days (no access)
Minimum object size for cost-effectiveness: 128 KB
No retrieval fees for Frequent, Infrequent, Archive Instant tiers
Deep Archive retrieval time: 12–48 hours standard
Storage classes: INTELLIGENT_TIERING (API value)
Edge Cases and Exceptions
If you access an object in the Deep Archive tier, you must first restore it using RestoreObject, which incurs retrieval fees and takes time. The object is not automatically promoted.
Intelligent-Tiering is not available in all AWS Regions (e.g., some China regions).
Objects in the Archive Instant tier are immediately accessible but stored on a lower-cost medium; there is no retrieval delay.
You cannot manually move an object between tiers within Intelligent-Tiering; S3 manages it automatically.
How to Eliminate Wrong Answers
If the scenario says "unpredictable access patterns" and the data is >128 KB, Intelligent-Tiering is likely the answer.
If the scenario says "data accessed frequently for 30 days then rarely," consider lifecycle policies to transition to Standard-IA after 30 days, but Intelligent-Tiering may also work.
If the scenario mentions "retrieval delays" or "instant access," note that only Deep Archive has delays; Archive Instant has no delay.
If the scenario mentions "monitoring fee," that's a clue for Intelligent-Tiering.
S3 Intelligent-Tiering automatically moves objects between four tiers based on last access time: Frequent (0-30 days), Infrequent (30-90 days), Archive Instant (90-180 days), Deep Archive (180+ days).
Monitoring fee is $0.0025 per 1,000 objects per month, regardless of size or access.
No retrieval fees for Frequent, Infrequent, or Archive Instant tiers; only Deep Archive incurs retrieval fees.
Objects smaller than 128 KB may not benefit from Intelligent-Tiering due to the per-object monitoring fee.
Accessing an object resets the clock and promotes it back to Frequent Access tier immediately (except Deep Archive requires restore).
Intelligent-Tiering is ideal for data with unknown or changing access patterns, such as user-generated content, media archives, and logs.
The Archive Instant tier provides immediate access with no retrieval delay, unlike S3 Glacier Flexible Retrieval.
Deep Archive tier requires a restore operation (12-48 hours) and incurs retrieval fees; it is not automatically promoted upon access.
Intelligent-Tiering is a storage class, not a lifecycle policy; you specify it at upload time.
You can combine Intelligent-Tiering with lifecycle policies to expire objects after a retention period.
S3 Intelligent-Tiering is not available in all AWS Regions; check regional availability.
Use S3 Storage Lens or S3 Inventory to monitor which tier objects are in.
These come up on the exam all the time. Here's how to tell them apart.
S3 Intelligent-Tiering
Automatic tier transitions based on access patterns, no manual rules.
Charges a monitoring fee per object ($0.0025/1000 objects/month).
No retrieval fees for Frequent, Infrequent, Archive Instant tiers.
No minimum storage duration charge.
Best for unpredictable or unknown access patterns.
S3 Lifecycle Policies (with Standard-IA/Glacier)
Requires manual lifecycle rules (e.g., transition to Standard-IA after 30 days).
No monitoring fee, but retrieval fees apply for Standard-IA and Glacier.
Standard-IA has a 30-day minimum storage charge; Glacier has minimum 90 days.
Lifecycle policies can also expire objects (delete).
Best for predictable access patterns (e.g., known infrequent access after 30 days).
Mistake
S3 Intelligent-Tiering charges retrieval fees for all tiers.
Correct
Only the Deep Archive Access tier incurs retrieval fees (per GB). The Frequent, Infrequent, and Archive Instant tiers have no retrieval fees.
Mistake
Intelligent-Tiering moves objects based on last modified date.
Correct
It moves objects based on last access date (last time the object was read via GetObject or HeadObject). The last modified date is irrelevant.
Mistake
Objects in Archive Instant tier are not immediately accessible and require restoration.
Correct
Archive Instant tier provides immediate access with no retrieval delay, just like Frequent and Infrequent tiers. Only Deep Archive requires restoration.
Mistake
Intelligent-Tiering is cheaper than S3 Standard for all objects.
Correct
For objects smaller than 128 KB, the monitoring fee ($0.0025 per 1,000 objects) may outweigh storage savings. S3 Standard may be more cost-effective for small objects.
Mistake
You can combine Intelligent-Tiering with lifecycle policies to manually control tier transitions.
Correct
Intelligent-Tiering is a storage class that handles transitions automatically. You cannot also apply lifecycle rules to the same object to force transitions—the storage class itself manages the tiers. However, you can use lifecycle policies to expire objects or transition them to Glacier after Intelligent-Tiering.
Reveal each answer, then mark whether you got it right. Score 60%+ to unlock the next chapter.
The monitoring fee is $0.0025 per 1,000 objects per month. This fee is charged for every object stored in the Intelligent-Tiering storage class, regardless of its current tier or access frequency. For example, storing 1 million objects costs $2.50 per month in monitoring fees.
The transition is not instantaneous. After the threshold is met (e.g., 30 days of no access), the move typically occurs within 24-48 hours. The exact timing is not guaranteed by AWS, but it is designed to happen within a couple of days.
Yes, S3 Intelligent-Tiering is compatible with S3 Object Lock in both governance and compliance modes. You can store locked objects in Intelligent-Tiering, and the automatic tier transitions will still occur. However, if the object is in Deep Archive tier and you need to access it, you must have the appropriate permissions to restore it.
Yes, you can apply lifecycle policies to objects stored in Intelligent-Tiering. For example, you can create a lifecycle rule that expires (deletes) objects after 365 days. However, you cannot use lifecycle policies to manually transition objects between Intelligent-Tiering's internal tiers; those transitions are automatic and managed by the storage class.
Unlike the other tiers, accessing an object in the Deep Archive tier does not automatically promote it. You must first initiate a restore operation using the RestoreObject API, which copies the object back to the Frequent Access tier (or you can specify a temporary copy). The restore takes 12-48 hours for standard retrieval and incurs retrieval fees. After restoration, the object is available for the duration you specify (default 1 day), after which the temporary copy expires.
Generally, no. The monitoring fee of $0.0025 per 1,000 objects is a fixed cost per object, not per GB. For small objects, the storage savings from moving to colder tiers may be less than the monitoring fee. AWS recommends using S3 Standard for objects smaller than 128 KB unless you have a very large number of objects and the monitoring fee is negligible.
Both provide immediate access to archived data, but they are different storage classes. S3 Intelligent-Tiering is an automatic tiering storage class that includes Archive Instant as one of its four tiers. S3 Glacier Instant Retrieval is a standalone storage class for data that is accessed rarely but requires millisecond retrieval. Intelligent-Tiering's Archive Instant tier has the same retrieval characteristics but is part of the automatic tiering system.
You've just covered S3 Intelligent-Tiering for Automated Cost Optimization — now see how well it sticks with free SAA-C03 practice questions. Full explanations included, no account needed.
Done with this chapter?