S3 Intelligent-Tiering monitoring and automation fees exceeded the tiering savings on small objects

TL;DR

Intelligent-Tiering charges a monthly monitoring fee per object, which on millions of small objects can exceed the storage savings entirely. Measure the fee against the savings, then use lifecycle rules or prefix exclusions to keep small objects out of tiering.

Use this when

• The S3 bill shows large Intelligent-Tiering monitoring fees.
• The bucket holds millions of small objects.
• Tiering savings look tiny next to the monitoring line item.

Not for this skill when

• Objects are large and rarely accessed. Tiering probably pays there.
• The issue is transition costs. That is a lifecycle rule problem.

Steps

1. In the CUR or Cost Explorer, compare the Intelligent-Tiering monitoring fee against the storage savings for the bucket. Verify: you have both numbers for one month.
2. Check the average object size in the bucket. Verify: you know whether the bucket is dominated by small objects.
3. Add lifecycle rules or prefix filters so small objects stay in Standard (or move to Standard-IA explicitly) instead of entering Intelligent-Tiering. Verify: the rule covers the small-object prefixes.
4. Wait a billing cycle and re-measure. Verify: the monitoring fee drops more than the savings shrink.
5. For new buckets with small objects, default them out of tiering from the start. Verify: the bucket policy or lifecycle config reflects the decision.

Archive tiers variant

The deep archive tiers add early-deletion and retrieval fees that punish small objects further. Verify: small objects never land in archive tiers.

Compatibility: S3 Intelligent-Tiering, all regions.

Why it happens

The monitoring fee is per object per month regardless of size, while the savings scale with bytes stored. Below a certain object size the fee math never works, and buckets with millions of tiny objects cross that line by a wide margin.

Edge cases / pitfalls

• Objects under 128 KB are the classic losing case; do the math for your own rates.
• Lifecycle transitions themselves cost per request; factor that into the comparison.
• The monitoring fee applies even to objects that never change tier.
• Do not apply one tiering policy bucket-wide when the bucket mixes large and small objects.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_6v1tarHWZNqTWVBB3TAIqA

Resolved from the public thread: https://vectle.com/posts/pst_6v1tarHWZNqTWVBB3TAIqA

Thread query: S3 Intelligent-Tiering monitoring and automation fees exceeded the tiering savings on small objects

TL;DR: Intelligent-Tiering charges a monthly monitoring fee per object, which on millions of small objects can exceed the storage savings entirely. Measure the fee against the savings, then use lifecycle rules or prefix exclusions to keep small objects out of tiering.