So I was reviewing the notes on EBS Snapshots and this is my interpretation...

The first snapshot is a full image of an ebs volume. Subsequent snapshots are incremental. This suggests that in a disaster recovery situation, you must have access to ALL of the snapshots taken to perform a recovery. i.e. You cannot, and should not, delete any of the snapshots, especially the first one taken. To recover when using incremental backups, you need all of the snapshots. To recover when using differential backups, you need the first and last snapshots to recover a volume.

http://aws.amazon.com... "Amazon EBS provides the ability to back up point-in-time snapshots of your data to Amazon S3 for durable recovery. Amazon EBS snapshots are differential backups, meaning that only the blocks on the device that have changed since your last snapshot will be incrementally saved. This means that if you have a device with 100 GBs of data, but only 5 GBs of data has changed since your last snapshot, only the 5 additional GBs of snapshot data will be stored back to Amazon S3."


Not really sure what to make of this because I don't want 50 snapshots laying around. It's confusing enough when they use differential and incremental in the same sentence.

Would someone kindly clarify which snapshots must be available to recover a volume in its entirety?

For the link that looks like what it does. I didn't realize that, but that is common for backups. I would assume you have to delete any previous snapshots and then create a new one in order to get a full snapshot.

One problem I have is I have 6 ebs volumes out there on our prod site and one snapshot. If you look at the snapshot you have no idea which volume it's from.

I think you may have misinterpreted what you read. Although internally Amazon stores incremental changes in the snapshot, you can safely delete earlier snapshots without losing any content from the more recent snapshots. Amazon will protect the information from earlier snapshots that are required to restore later ones. This is all transparent to the user.

Just getting started with EBS snapshots and wondering if someone can point me to a good tutorial.

I intend to set up a cron job to take a nightly snapshot of my EBS volumes.

Once I've done that, how do I restore files from those snapshots? Can I restore individual files? Do I have to mount the snapshot as another drive and copy the desired files back over? What's the best practice for using EBS snapshots as a backup mechanism? Or is the best practice NOT to use them as a backup mechanism and to use something else instead?

Thanks!

Remember that an EBS snapshot is a copy of the filesystem at an instant in time. You do not have any direct access to the snapshot content except through recreating the filesystem in EBS where it can be mounted and individual files copied over. We have used EBS this way successfully but only rarely need to restore files. This approach seems to work well at the small scale.

If you need to do this frequently or at a larger scale then it might be better to invest in backup software to manage this (products like Amanda support backups to S3 and there are probably other products as well).

This is correct. A snapshot is complete in itself.

This is still not working out so well. I have 200+ snapshots going back to December that will not delete because it says they are in use by ami-nnnnn. Unfortunately ami-nnnnn is long gone from the account.... so I cannot perform a detach/dismount.

Mike, these sound like they are EBS volumes, not snapshots. If they are volumes and the AMI is gone then I'd recommend that you report this on the Amazon AWS forum - they should be able to resolve it. I've seen issues like this before - way back when the were just introducing EBS but have not heard of it since.