upload S3 backups using different storage classes
Currently, it appears as if the S3 backup mechanism will use the default S3 Standard storage class, but it would be helpful to allow selection of Standard-Infrequent Access (IA) and Reduced Redundancy storage classes as well. This would allow customers to better control AWS costs and service level objectives associated with storage of backups.
Isn't this something that can be handled via S3 lifecycle rules on the AWS side?0
No, this is not the same as lifecycle policies. Lifecycle policies can transition objects from one storage class to another and even delete objects after a certain period of time from the initial upload.
At the time of upload, one can specify the storage class for the object other than the default STANDARD class. If using the AWS CLI, this is controlled by the '--storage-class' flag.
The STANDARD storage class doesn't make sense for most archival or backup data because when retrieving and restoring from backup, time to first byte (TTFB) latency of the download is not that important—only the immediate availability of the backup data is important. Backups are also not intended to be accessed frequently. Besides that, the STANDARD storage class is almost twice as expensive as STANDARD_IA.
The Reduced Redundancy class is more expensive than the Standard class at this time. It shouldn't be used, and AWS are planning to deprecate that storage class entirely per https://mysteriouscode.io/blog/aws-s3-storage-classes-pricing-is-not-what-you-think/#update10
Please sign in to leave a comment.