Today I reviewed a client case involving migration of multiple petabytes between AWS accounts. On paper — S3 cross-account copies, what could go wrong? Turns out, everything.
Where's the catch?
Key pitfalls:
- IAM policies on source buckets don't carry over
- Object Lock requires special handling
- Transfer Acceleration can spike costs without proper rate limits
- Initial snapshot via Storage Gateway
- Delta sync using S3 Batch Operations
- Final validation with Athena checksums
Has anyone faced similar challenges? How did you handle large-scale cross-account migrations?
What saved the day
A hybrid approach worked: