Practical guide
S3 small files: why object count and request patterns matter
estimate S3 request costs for many small files. Includes a worked example, calculator scope, and practical questions.
Updated · Sources checked
Equal stored bytes can produce very different request counts
A million small objects and a few large objects can occupy similar storage but require different numbers of operations to write and retrieve. Model request activity independently from total GB before concluding that two data layouts cost the same.
This is particularly useful for exports, event archives, and build artifacts. The right layout still depends on access patterns and recovery needs, so fewer objects is not automatically the best application design.
Worked example: daily event files
Suppose a pipeline writes one million objects per day. In a hypothetical 30-day month, a one-request-per-object write path generates 30 million write requests.
An alternative batches the same records into 10,000 objects per day. Under the same simplified write assumption, it generates 300,000 monthly writes. The difference is 29.7 million write requests. Storage bytes remain an explicit separate assumption; file headers, compression, and indexing can change them.
Use the S3 calculator to compare the supported write-request component at the same average storage input. Do not report the request reduction as an equal percentage reduction in the total bucket bill.
Include how readers use the data
If a user needs one small record, a large bundled object may require retrieving more data or doing additional processing. Batching can also add latency before records become available. Estimate read operations, bytes transferred, and compute work for both layouts before choosing one.
Count actual API operations from the pipeline. Multipart upload behavior, metadata checks, listing, retries, and cleanup can make the real request count differ from the simple one-write-per-object scenario. Use the calculator only for the request categories it explicitly supports.
The tool covers first-tier S3 Standard storage and common requests in US East. It excludes transfer, other classes, lifecycle transitions, replication, and additional services. If the proposal involves archival classes, minimum object-size and retention rules need a separate evaluation as well.
Frequently asked questions
Do fewer objects always mean less storage? No. Compare actual stored bytes; batching changes the layout, not necessarily the logical data size.
Can I claim 99% total savings from 99% fewer writes? No. Only the affected write component follows that arithmetic, while storage, reads, delivery, and processing can remain or increase.
Calculate your scenario
Use the s3 cost calculator. Keep the displayed region, pricing date, billing units, and exclusions alongside your result. The arithmetic examples above illustrate usage or explicitly hypothetical rates; they are not AWS quotes.
Source
Billing structure checked against official AWS documentation on September 6, 2026.