---
title: "Comment by CDN Edge Ops on \"CloudFront Overview and CloudFront with S3: A Detailed Guide\" | Hashnode"
description: "Putting S3 behind CloudFront is straightforward; keeping origin request rate down is mostly cache-key and header discipline. Prefer a REST or website origin wit"
url: "https://hashnode.com/posts/cloudfront-overview-and-cloudfront-with-s3-a-detailed-guide/66c6c443c9ac2057c9a75a92/comment/6aec5595cf2826f165801f6a"
---

[Discussion](https://hashnode.com/posts/cloudfront-overview-and-cloudfront-with-s3-a-detailed-guide/66c6c443c9ac2057c9a75a92)

Comment [CloudFront Overview and CloudFront with S3: A Detailed Guide](https://hashnode.com/posts/cloudfront-overview-and-cloudfront-with-s3-a-detailed-guide/66c6c443c9ac2057c9a75a92)

[CDN Edge Ops](https://hashnode.com/@cdnedgeops)

[Nov 5](https://hashnode.com/posts/cloudfront-overview-and-cloudfront-with-s3-a-detailed-guide/66c6c443c9ac2057c9a75a92/comment/6aec5595cf2826f165801f6a)

Putting S3 behind CloudFront is straightforward; keeping origin request rate down is mostly cache-key and header discipline. Prefer a REST or website origin with an origin access control that only the distribution can call, emit Cache-Control from the object (or a response-headers policy) so edges can store hashed assets long-term, and avoid forwarding the full query string or Cookie on the default behaviour — each unique variant becomes its own object and Origin Shield cannot collapse what the key already split.

Ops prove-out after wiring the origin and cache policy: hit a hashed object twice and expect MISS then HIT with a rising Age, then fetch the HTML shell and confirm it stays on a short TTL or purge tag. After replacing an object in the bucket under the same key, run an invalidation for that path (or rely on versioned filenames) and re-check X-Cache until the old body is gone everywhere that mattered — most “S3 update not visible” tickets are an un-invalidated edge object or a query variant still in the key, not a broken bucket ACL.
