Why it matters
Many buckets hold data written once and rarely read again, but pay the S3 Standard price for all of it.How nOps finds it
nOps reads the last 90 days of S3 billing, including the object count from S3 Inventory. A bucket is flagged when it has at least $5 of Standard storage in the last 30 days, has an S3 Inventory report from the last 45 days, held only Standard storage with no lifecycle transitions for 90 days, and was billed for Standard storage on at least 85 of those days. Objects must average at least 128 KB. The share of objects not accessed is at least one minus object-level requests divided by the object count, and must be at least 50%. The saving is the Standard storage cost times that share times the price difference between the Frequent and Infrequent Access tiers, minus the Intelligent-Tiering monitoring fee, and must be at least $25 a month.Before you start
S3 Inventory enabled on the bucket (nOps needs its object count), and permission to update lifecycle rules (s3:GetLifecycleConfiguration, s3:PutLifecycleConfiguration).
How to implement
- Save the bucket’s current lifecycle rules:
aws s3api get-bucket-lifecycle-configuration --bucket BUCKET. - Add a rule that transitions objects to
INTELLIGENT_TIERINGafter 0 days, keeping every existing rule.put-bucket-lifecycle-configurationreplaces all rules, so write back the full merged configuration. - In the console: S3 → bucket → Management → Create lifecycle rule → Move current versions of objects between storage classes → Intelligent-Tiering.
- Update Terraform or CloudFormation with the same rule.
- Check after about a month that Intelligent-Tiering storage appears on the bill.
Risks and considerations
- Savings start after objects go 30 days without access.
- Intelligent-Tiering charges a small monitoring fee per object, which is why small objects are excluded.
- The share is counted by objects, not bytes, so the real saving can differ.
- To roll back, restore the saved lifecycle rules. Objects already in Intelligent-Tiering stay there, with no retrieval fees.
FAQs
Why does nOps need S3 Inventory?
Why does nOps need S3 Inventory?
Billing shows requests but not how many objects a bucket holds. The Inventory object count lets nOps work out how many objects were not touched.