Confused about archiving and restoring using Glacier Instant Retrieval

Question

I was concerned about inadvertent emptying and deleting of my S3 bucket and so decided to make an archive copy. Glacier Instant Retrieval seemed like the right candidate. So, I set up the replication rule on my bucket and replicated the existing objects.

To test restoration, I deleted an object from my bucket. Now I want to initiate restore from the replica, but the Initiate Restore action is grayed out. When I run `aws s3api restore-object` in CLI, I get the error: `An error occurred (InvalidObjectState) when calling the RestoreObject operation: Restore is not allowed for the object's current storage class`. What am I missing?

Another thing that's not clear to me is about Glacier Instant Retrieval being designed to be accessed once in a quarter. Does this mean that if I restore my bucket once and another emergency happens less than 90 days from then, I won't be able to restore my bucket?

Accepted Answer

Hi there,

The Amazon S3 Glacier Instant Retrieval storage class doesn't require to restore objects. You can simply download the objects like you would do with the Amazon S3 Standard storage class. Be aware that when you download objects in the Glacier Instant Retrieval storage class, you will be charged data retrieval fees per GB (i.e. in the AWS region us-east-1 you would pay $0.03 per GB).
Because of Glacier Instant Retrieval being cheaper in terms of storage costs but more expensive to download, we recommend to use Glacier Instant Retrieval no more than once in a quarter. Technically you can access data sets as many times as you want but, if you do so more than ~4 times per year, you would pay more for Glacier Instant Retrieval than for S3 Standard.

See [Amazon S3 pricing](https://aws.amazon.com/s3/pricing/) for more information.

Confused about archiving and restoring using Glacier Instant Retrieval

Contenuto pertinente