Naturally, the KVM backup files have a timestamp in their name. Will dedplication also work for files with different names? Or rather, is deduplication working on a repo-scale or only file-wise?
Kopia chops up files into chunks and these chunks get stored only once, i.e., deduplicated. In your example, everything the two backup files have in common will only get stored once, irrespective of the file names or location.
The documentation provides additional details: https://kopia.io/docs/architecture/
Thanks, that’s what I wanted to hear.