Amazon EC2: Moving an EBS volume snapshot to s3

Question

I am trying to create a backup and restore mechanism for my EBS volumes

I could easily create a mechanism for this by using

to take a snapshot of volume

ec2-create-snapshot -d $description $volume

and restore

ec2-create-volume --snapshot $snapid -z $zone

Now as per my understanding, these snapshots are still on EBS. So a better way will be to move the snapshots to S3.

So how can I move the snapshots to s3 bucket?

Solution 1

EBS snapshots are stored in S3. There should not be any need to move the snapshot anywhere.

Solution 2

The question seems sensible enough to me. In our case we want to take an EBS volume that we are using in us-west-1 region and provide an image of it (via S3) so that we can attach it to an instance running in us-east-1.

As simple as this concept seems, there is not apparently a simple solution currently, nor even a solution at all. If anybody knows of one please post info or links here, thanks.

Comments

[FlavorScape]

I don't see them. does it create a default bucket? don't seee that either.

[Mike Brant]

@FlavorScape EBS snapshots are not in any bucket that you own. The persistence layer is S3 however. The snapshots in fact are not even complete snapshots of data, but rather incremental updates from previous snapshots, so even if you as an end user had access to the raw binary snapshot (which you don't) you would have to know how to assemble the pieces of all previous snapshots to get something usable. You just work with snapshots based on their ID and either mount volumes via AWS console, or via command line tools.

[Mike Brant]

You CAN copy snapshots across regions. See the documentation here docs.aws.amazon.com/AWSEC2/latest/UserGuide/…

[Octopus]

@ Mike, yes. As nnahum said, there is a new feature. It was released late in Dec 2012 that allows us to do this now :)

[Devplex]

I'm not sure I understand. Snapshot copies are still hidden from your S3 buckets. That means I cannot manage them like actual S3 objects. Am I misreading?

[Adam Ocsvari]

The need for moving snapshots is the storage costs. It is significantly cheaper to store in S3 or in Glacier compare to snapshots.

[Mike Brant]

@AdamOcsvari EBS snapshots are physically stored in S3. You are correct that pricing for EBS snapshots is (for whatever reason) on the order of 3X the cost of standard S3 storage. However, say you were saving 1TB of data and between snapshots you change (but not grow) ~100GB of data. With EBS snapshots, you will still only be paying for 1TB no matter how many snapshots you make. If you were to store the entire data set in S3 in a non-incremental way, you would only be able to store ~ 3 full "snapshots" of the data for the same price it costs to store the data set in EBS.

[Paulo Neves]

An answer saying the question is unnecessary is not an answer. The question might be unsafe but this answer does not answer the question of how to move snapshots to an S3 bucket.