Ops Manager currently uses S3 Standard storage for all snapshot storage and oplog storage and the storage class is not configurable.

AWS has storage classes that are specifically designed for backups (lower storage costs, but higher retrieval costs) like GlacierInstantRetrieval.

We could significantly reduce costs if we could select the Storage Class that Ops Manager would use when uploading data to S3.

NOTES

It is possible to create lifecycle rules in S3 to transition the objects from Standard Storage to another Storage Class, but this transition itself incurs a cost for the objects/bytes that are transitioned. This may reduce costs some, but not as much as if the objects used the desired Storage Class during creation.

It is NOT possible to set a "default storage class" on an s3 bucket.