In Dec 2020, Amazon S3 shifted to strong read-after-write globally — did anyone else delete a bunch of cache-warming and retry jitter hacks after that? We dropped about 30% of idempotency wrappers in a cross-account ingestion pipeline and saw cleaner multi-region fan-in with more predictable scale-up.
“pipeline and saw cleaner multi-region fan-in with more predictable scale-up.” Same here — after the 2020 flip (Amazon S3 Update – Strong Read-After-Write Consistency | AWS News Blog), we killed our 2–5s GET-after-PUT jitter and just do a single GET (or HEAD+GET with VersionId when we have it), which trimmed about 20% off ingest latency; only caveat: we still keep a tiny retry for cross-account KMS decrypts. Did you also drop SQS visibility padding once you trusted strong R-A-W?