A backup is a stored copy of data used to recover after loss, corruption, accidental deletion, or disaster. Keeping a copy offsite reduces the risk that one event destroys both the original and its recovery data. AWS provides several approaches: S3 stores objects, EBS provides durable block storage, and managed databases provide service-specific recovery features. section 7.3 section 8.2.3
A simple S3 backup and restore workflow is:
- Copy a local directory into a backup prefix with `aws s3 sync`; the command uploads only new or changed files. section 7.3
Choose the storage method according to the workload. Instance store volumes have no built-in backup mechanism, so scheduled jobs can periodically synchronize their data to S3; for data requiring durable block-level storage, EBS is the better choice. RDS supplies on-demand snapshots and point-in-time restore within a retention period of up to 35 days. DynamoDB replicates data for durability but still benefits from on-demand snapshots that recover from accidental deletion or application-caused corruption. section 8.2.3 section 10.8 section 12.10
Recovering a web application's files
A web application can synchronize its local backup directory to an S3 bucket, then restore the contents into a new local directory when the original files are lost. Keep the bucket name unique, avoid unexpectedly large test uploads, and account for storage and transfer charges. section 7.3
