A snapshot is an image of the entire storage volume for your database, not just a single instance as you get with a native database backup. If you have a multi-AZ setup, as described in a previous recipe, there is no I/O cost associated with the creation of the snapshot. Snapshots are stored in S3, so you get the built-in durability of S3 buckets for your snapshot. You give the snapshot a unique identifier so that you can reference it if you need to restore the snapshot later.