Each EBS snapshot only gradually adds blocks that have changed since the last snapshot.
Each EBS snapshot has all the blocks that were ever used on an EBS volume. You can delete any snapshot without reducing the completeness of any other snapshot.
This is magic.
Well, actually this is a bit of a technological focus, when each shot has pointers to blocks that it likes, and several shots can share the same blocks. As long as there is at least one snapshot that points to a specific data set on the block, the block is stored on S3.
This prevents Amazon from telling you how much space a single shot takes, because their sizes are not mutually exclusive.
Here's an old RightScale article that has some pretty pictures explaining how snapshots work behind the scenes:
http://blog.rightscale.com/2008/08/20/amazon-ebs-explained/
Note that snapshots only save blocks on the EBS volume that were used, and snapshots are compressed, which further reduces the cost of storing data.
Eric hammond
source share