The Truth About AWS EBS Snapshots

EBS Snapshots Are Physically Stored in S3 Behind the Scenes

Amazon EBS volumes can be snapshotted for backup and disaster recovery purposes. When an EBS snapshot is taken, the volume data is copied and stored in Amazon S3. However, EBS snapshots do not appear as normal S3 objects that users can access through their S3 console or API.

Snapshots Are Stored in a Hidden S3 Bucket

While EBS snapshots leverage the durability and scale of S3 for storage, the S3 bucket used to store snapshot data is not visible to the end user. AWS manages this bucket internally to store the snapshot objects. From a customer perspective, snapshots only appear as resources that can be managed through the EC2 console or API under the “Snapshots” tab of EBS volumes. This hidden S3 bucket approach simplifies snapshot management for users while leveraging S3’s capabilities “under the hood.”

Transferring Snapshots Across Regions Through S3

While snapshots are stored in S3, the default behavior is that they are only accessible from the original region where the volume resided. If a customer wanted to copy a snapshot to another region for deployment or disaster recovery purposes, they would traditionally use the “Copy Snapshot” feature in EC2.

Digging Into How Copy Snapshot Works

Upon further investigation, it was discovered that the “Copy Snapshot” process actually leverages S3 cross-region replication behind the scenes. When a snapshot is copied, the snapshot data is replicated from the source S3 bucket to the S3 bucket in the target region. From the user perspective, this process results in an independent snapshot being created in the other Availability Zone that appears normally under the “Snapshots” tab.

Leveraging S3 Capabilities Without Complexity

By performing the cross-region transfer at the S3 layer invisibly, AWS is able to leverage the scalability and features of S3 without increasing complexity for the end user. Customers benefit from the ability to copy snapshots across regions while still interacting with snapshots as simple EC2 resources.

The trade-off of Storing Snapshots in S3 Buckets

While storing EBS snapshots in S3 provides benefits, there are some trade-offs to consider as well. Placing snapshot objects directly in customer S3 buckets could allow for some potential advantages but would require more complex snapshot management.

Potential Upsides of Customer S3 Bucket Storage

Storing snapshots in customer S3 buckets could allow leveraging cheaper S3 Infrequent Access or Glacier storage tiers. It may also simplify building cross-account disaster recovery solutions by pushing snapshot copies to other accounts.

Drawbacks of More Visible S3 Storage

However, compiling fragmented incremental snapshot objects from S3 into a consistent volume would become very challenging. Additional API calls and latency would be introduced to assemble snapshots on demand. More complex plumbing would be needed to maintain the seamless user experience of snapshot management that AWS currently provides. Overall, directly exposing snapshots as S3 objects comes with certain trade-offs compared to the more opaque “snapshot as a resource” model used today. Both approaches have merits depending on specific use cases and priorities.

Conclusion

In summary, while EBS snapshots are physically stored in S3, AWS abstracts away these underlying details to provide a simplified user experience. Leveraging S3 capabilities “behind the curtains” allows snapshots to be easily managed across regions without overcomplicating the user interface. While alternate models exist, AWS’s approach balances simplicity and leveraging scalable cloud infrastructure to deliver cost-effective, simple-to-use snapshot functionality for blockchain workloads.