Is continuous backup of S3 bucket-wise, file-wise or byte-wise?

Question

Assume I have an S3 bucket with continuous backup enabled. Now there are some file a in it, which is heavily updated.

How do we do the continuous backup?

* Is it bucket-wise? For every change per file, we have a copy of the entire bucket?
* Is it file-wise? For every change per file, we have a new copy of the entire file?
* Is it byte-wise? For every change per fille, we just store the updates of the individual bytes?

Personally, since there are many small changes per file and the files can by quite considerable in size, I would prefer to do something like the byte-wise solution.

Answer

Amazon S3 objects cannot be edited. Objects can be read (with GET), and updated versions written (with PUT). The whole object is always written, and there is no tracking of byte-level differences between objects or versions of object keys. 
Therefore, ongoing backups of a bucket will contain the complete data in all new objects PUT since the last backup. This is similar to your 'file-wise' analogy.

https://docs.aws.amazon.com/aws-backup/latest/devguide/s3-backups.html
> For both backup types, the first backup is a full backup, while subsequent backups are incremental at object-level.

Answer

Hello.

S3's Continuous Backups are incremental backups, so I think they create backups of parts where changes occur.  
For example, if a 1 GB object has a 1 KB change, subsequent backups will create a new 1 GB object in the backup vault.  
https://docs.aws.amazon.com/aws-backup/latest/devguide/s3-backups.html  
> For both backup types, the first backup is a full backup, while subsequent backups are incremental at object-level.

Is continuous backup of S3 bucket-wise, file-wise or byte-wise?

Relevant content