Do incremental backups read each file or is it just a stat comparing time and size to see if the file has changed?

Is there a way to make it only read time and size differences and if there is a difference only then read the file?

Judging on the speed of my incremental backups, its doing exactly that. I don’t know if it only looks at size and mtime, or other factors, but it is definitely not reading the entire file for incremental backups.

I think file changes are determined by comparing hashes, which is more reliable than size/timestamp/attributes.

It depends on the OS. On Windows it uses the USN + keeps track of open files. On Linux it uses mtime+ctime (plus size in both cases). If it uses CBT it skips unchanged parts of files and doesn’t read them.

1 Like