S3 storage classes compared: when Standard-IA, Intelligent-Tiering and Glacier actually save money
S3 Standard-IA costs $0.0125 per GB-month against $0.023 for S3 Standard in US East (N. Virginia) (AWS S3 pricing page, checked November 2026), but the move saves money only when a bucket's objects are larger than 128 KB, stay past the 30-day minimum storage duration, and are rarely read. A bucket of small, short-lived or frequently read objects can cost you more in Standard-IA or Glacier than it did in S3 Standard, so check those three things before you write a lifecycle rule.
Why does my S3 bill go up after I enable lifecycle rules?
A lifecycle rule to Standard-IA or Glacier raises the S3 bill when the side costs outweigh the storage saving. The side costs are objects under 128 KB billed as 128 KB, one transition request per object, retrieval and GET fees on every read, and early deletions billed for the full minimum duration.
The fees that reverse the saving, in US East (N. Virginia) (AWS S3 pricing page, checked November 2026):
| Class | Storage per GB-month | Retrieval per GB | GET per 10,000 |
|---|---|---|---|
| S3 Standard (first 50 TB) | $0.023 | none | $0.004 |
| Standard-IA | $0.0125 | $0.01 | $0.01 |
| One Zone-IA | $0.01 | $0.01 | $0.01 |
| Glacier Instant Retrieval | $0.004 | $0.03 | $0.1 |
Glacier Flexible Retrieval ($0.0036 per GB-month) and Glacier Deep Archive ($0.00099 per GB-month) objects must be restored before they can be read. A Glacier Flexible Retrieval restore costs $0.03 per GB plus $0.01 per request (Expedited), $0.01 per GB plus $0.050 per 1,000 requests (Standard), or $0.00 per GB plus $0.00 per 1,000 requests (Bulk). Every restore also bills a temporary copy at the S3 Standard rate, on top of the archived object.
How do I check if my objects are the right size?
Turn on S3 Inventory for the bucket (S3 console > bucket > Management > Inventory configurations) with the Size, Last modified date and Storage class fields, and query the report in Athena. Inventory is delivered daily or weekly, not on demand, so set it up now. Then look at the size distribution by count and by bytes:
SELECT
CASE WHEN size < 131072 THEN 'under 128 KB' ELSE '128 KB and over' END AS size_band,
COUNT(*) AS objects,
SUM(size) AS bytes
FROM s3_inventory_table
WHERE bucket = 'your-bucket-name'
AND dt = '2026-11-12-01-00' -- latest inventory snapshot
GROUP BY 1
Standard-IA, One Zone-IA and Glacier Instant Retrieval bill any object under 128 KB as 128 KB. Glacier Flexible Retrieval and Glacier Deep Archive add 40 KB to every archived object: 32 KB at the Glacier class rate plus 8 KB at the S3 Standard rate (AWS S3 user guide, checked November 2026).
Since September 2024, S3 Lifecycle by default does not transition objects smaller than 128 KB, because each object is one transition request and "the transition costs can outweigh the storage savings." Read the query by bytes, not only by count. The under-128 KB band is the data that costs more after a move, and the bytes in the other band are all a colder class can save.
How do I check if objects stay past the minimum duration?
Inventory shows how many of today's objects have already outlived each minimum. Run this on the bucket while it is still in Standard:
SELECT
COUNT(*) AS objects,
SUM(CASE WHEN last_modified_date < current_timestamp - interval '30' day THEN 1 ELSE 0 END) AS older_than_30d,
SUM(CASE WHEN last_modified_date < current_timestamp - interval '90' day THEN 1 ELSE 0 END) AS older_than_90d,
SUM(CASE WHEN last_modified_date < current_timestamp - interval '180' day THEN 1 ELSE 0 END) AS older_than_180d
FROM s3_inventory_table
WHERE bucket = 'your-bucket-name'
AND dt = '2026-11-12-01-00' -- latest inventory snapshot
Standard and Intelligent-Tiering have no minimum storage duration. Standard-IA and One Zone-IA have 30 days, Glacier Instant Retrieval and Glacier Flexible Retrieval 90 days, and Glacier Deep Archive 180 days. Deleting or overwriting an object before its minimum is billed for the full minimum; for archived objects this is a prorated early deletion fee. Compare those minimums with your expiration rules and with how often the application overwrites keys.
If data is already in a colder class, open Cost Explorer, filter Service to S3, group by Usage type and look for usage types containing EarlyDelete. Spend there means objects are leaving before their minimum.
How do I check my retrieval rate?
Measure reads with S3 CloudWatch request metrics. They are off by default, count only from the day you enable them, and are billed as CloudWatch metrics. Compare a month of BytesDownloaded with the bucket's size.
Compare per GB. Standard-IA stores for $0.0125 per GB-month against $0.023 for S3 Standard and charges $0.01 per GB retrieved, so reading the bucket's full size about once a month cancels the saving. Glacier Instant Retrieval stores for $0.004 per GB-month and charges $0.03 per GB retrieved, so one full read costs more than a month of what the move from Standard saves.
Enable request metrics for the whole bucket, then, a few weeks later, pull daily bytes downloaded:
# Today: enable request metrics for the whole bucket
aws s3api put-bucket-metrics-configuration \
--bucket your-bucket-name \
--id EntireBucket \
--metrics-configuration '{"Id": "EntireBucket"}'
# A few weeks later: daily bytes downloaded since you enabled them
aws cloudwatch get-metric-statistics \
--namespace AWS/S3 \
--metric-name BytesDownloaded \
--dimensions Name=BucketName,Value=your-bucket-name Name=FilterId,Value=EntireBucket \
--start-time "$(date -u -d '28 days ago' +%Y-%m-%dT00:00:00Z)" \
--end-time "$(date -u +%Y-%m-%dT00:00:00Z)" \
--period 86400 \
--statistics Sum
For data already in a colder class, the Cost Explorer usage types containing Retrieval show the retrieval fees you pay today.
Check your own bill in 2 minutes: https://revelations.consulting/checker?utm_source=blog&utm_medium=organic&utm_content=blog-s3-storage-classes-compared
How do I fix it without guessing?
Start with S3 Intelligent-Tiering when you don't know the access pattern. In US East (N. Virginia), Intelligent-Tiering charges $0.0025 per 1,000 objects per month for monitoring and automation and has no retrieval fees (AWS S3 pricing page, checked November 2026). It moves an object to Infrequent Access after 30 consecutive days without access, to Archive Instant Access after 90, and back to Frequent Access when the object is read. Objects under 128 KB are not monitored and stay in Frequent Access, so they never pay a 128 KB minimum.
Use a lifecycle rule to a specific class only when the access pattern is predictable and the bucket passes all three checks: objects above 128 KB, kept past the class's minimum duration, and read rarely. Either route has a one-time cost. In US East (N. Virginia), lifecycle transitions are $0.01 per 1,000 requests to Standard-IA, One Zone-IA or Intelligent-Tiering, $0.02 to Glacier Instant Retrieval, $0.03 to Glacier Flexible Retrieval and $0.05 to Glacier Deep Archive, one request per object (AWS S3 pricing page, checked November 2026). Moves between tiers inside Intelligent-Tiering carry no transition fee.
Scope the first rule to one prefix with an object-size filter; transitions only go down. put-bucket-lifecycle-configuration replaces the bucket's entire lifecycle configuration, including expiration rules. Export the current rules with aws s3api get-bucket-lifecycle-configuration --bucket your-bucket-name and merge into them:
{
"Rules": [
{
"ID": "logs-to-standard-ia",
"Status": "Enabled",
"Filter": { "And": { "Prefix": "logs/", "ObjectSizeGreaterThan": 131072 } },
"Transitions": [{ "Days": 30, "StorageClass": "STANDARD_IA" }]
},
{
"ID": "abort-incomplete-uploads",
"Status": "Enabled",
"Filter": { "Prefix": "" },
"AbortIncompleteMultipartUpload": { "DaysAfterInitiation": 7 }
}
]
}
Glacier Deep Archive stores for $0.00099 per GB-month in US East (N. Virginia). A restore costs $0.02 per GB plus $0.10 per 1,000 requests (Standard) or $0.0025 per GB plus $0.025 per 1,000 requests (Bulk) (AWS S3 pricing page, checked November 2026). The Deep Archive rate, the 40 KB overhead and the 180-day minimum all start the day the lifecycle rule is satisfied, even before the physical move. Deep Archive fits large objects you keep past 180 days and rarely restore.
What it usually looks like
The bill rises the month after a new lifecycle rule runs. In Cost Explorer, filter Service to S3 and group by Usage type. Storage drops, but request charges jump that month, one transition request per object, and usage types containing EarlyDelete or Retrieval appear in the months after.
The size filter was set below 128 KB. A rule with ObjectSizeGreaterThan under 131072 overrides the September 2024 default, and every small object in Standard-IA, One Zone-IA or Glacier Instant Retrieval is billed as 128 KB. Check the Filter block in get-bucket-lifecycle-configuration.
An expiration rule ends before the destination class's minimum. Objects moved to Glacier Flexible Retrieval or Deep Archive and then expired early are billed for the rest of the 90 or 180 days. Compare each rule's Expiration days with its Transitions.
Archived data still gets restored. Restore and retrieval usage types show up every month, along with S3 Standard storage for the temporary restored copies.
The bucket keeps every version and every failed upload. aws s3api list-multipart-uploads --bucket your-bucket-name lists incomplete uploads, and an Inventory report that includes all versions shows noncurrent ones where is_latest is false. AbortIncompleteMultipartUpload is safe to add. NoncurrentVersionExpiration permanently deletes old versions, so confirm nobody relies on them for recovery or compliance before you add it.
Before you write the rule, run the payback test. Divide the one-time lifecycle transition cost (number of objects ÷ 1,000 × the transition price per 1,000 requests) by the monthly storage saving. If the result is longer than the objects will live, leave them where they are. If you want a second pair of eyes on the numbers, book a free 30-minute consult: we'll look at your bill with you, no obligation and no sales pitch.
Run your own numbers: S3 storage class calculator
Frequently asked questions
What is the minimum storage duration for S3 Standard-IA?
S3 Standard-IA has a 30-day minimum storage duration. Deleting or overwriting an object in Standard-IA before 30 days is billed as if it had stayed the full 30 days (AWS S3 user guide, checked November 2026).
Why does Intelligent-Tiering not charge retrieval fees?
S3 Intelligent-Tiering charges a monitoring and automation fee of $0.0025 per 1,000 objects per month in US East (N. Virginia) instead of retrieval fees, and moves an accessed object back to Frequent Access. Objects in the optional Archive Access and Deep Archive Access tiers must be restored before they can be read (AWS S3 pricing page, checked November 2026).
What is the cost of transitioning an object to Glacier Deep Archive?
A lifecycle transition to S3 Glacier Deep Archive costs $0.05 per 1,000 transition requests in US East (N. Virginia), one request per object (AWS S3 pricing page, checked November 2026). From then on, each object is billed the 40 KB overhead and held to the 180-day minimum storage duration.
Are objects under 128 KB monitored by Intelligent-Tiering?
No. S3 Intelligent-Tiering does not monitor objects smaller than 128 KB: they always stay in the Frequent Access tier, never move to Infrequent Access or the archive tiers, and are not charged the monitoring and automation fee (AWS S3 user guide, checked November 2026).
Run the numbers on your own bill: What would this bucket cost in another S3 storage class?. Free, in your browser.
Want this kind of thing weekly? Get Cloud Bill Weekly. Want us to find it in your bill? We're taking three founding clients for a free Cost Assessment.