Is it possible to cross-region replicate S3 data in Intelligent Tiering without paying for 30 days of Frequent Access?

Question

I have a bucket with hundreds of terabytes of data in a bucket in Intelligent Tiering, and would like to have a cross-region DR backup to Glacier. Using Batch Replication to create the initial version of that backup bucket would transition all objects to Frequent Access tier, and I'd pay for all objects at that tier for 30 days before some could transition down to Infrequent Access.

Is there any way to start a cross-region Glacier backup without paying that 30 days of Frequent Access? My read of the docs right now is that there is not, and the 30 day cost is unavoidable.

Answer

Hi,

I agree with your assessment. Copying the object either manually, via Replication or Batch Operations will tier up the object (see https://docs.aws.amazon.com/AmazonS3/latest/userguide/intelligent-tiering-overview.html)

> The following actions constitute access that automatically moves objects from the Infrequent Access tier or the Archive Instant Access tier back to the Frequent Access tier:
> 
>   Downloading or copying an object through the Amazon S3 console.
> 
>   Invoking CopyObject, UploadPartCopy, or replicating objects with Batch Replication. In these cases, the source objects of the copy or replication operations are tiered up.
> 
>   Invoking GetObject, PutObject, RestoreObject, CompleteMultipartUpload, or ListParts.

AWS Backup calls this out explicitly, see https://docs.aws.amazon.com/aws-backup/latest/devguide/s3-backups.html

> If you create a backup of an S3 Intelligent Tier (INT) object, then the source object moves to a storage tier that is more expensive than its present storage tier.

Is it possible to cross-region replicate S3 data in Intelligent Tiering without paying for 30 days of Frequent Access?

Relevant content