In T1372#26104, @ftigeot wrote:

Borgbackup is unable to pull data from remote hosts to a central location.

I do not understand this assertion. I mean (if this is the point) do we really care whether it is a push or pull based model?

See:

https://borgbackup.readthedocs.io/en/stable/quickstart.html#remote-repositories

and

https://borgbackup.readthedocs.io/en/stable/quickstart.html#important-note-about-free-space

There are some interesting comments on https://www.reddit.com/r/linux/comments/42feqz/i_asked_here_for_the_optimal_backup_solution_and

Do not loose the objective of this ticket: we want a (rather) complete comparison between these solutions, with references to existing blog posts/articles, pros/cons, etc. Ideally with some results from experiments on not too big but realistic data sets. It could even take the form of a public blog post!

Borgbackup is unable to pull data from remote hosts to a central location.

I do not understand this assertion.

This is the way Borgbackup was made to work. The authors simply don't want to have it pull data from remote servers but instead process it locally.
There are ways and hacks to provide remote data transfers but the result is not optimal. At the very least, there will be performance losses compared to what other solutions are capable of doing.

Comparison between Backuppc and Rsnapshot done, now adding Restic - https://restic.net/ - to the mix.
Borgbackup not tested yet.

Attachment: comparison between Backuppc and Rsnapshot

BorgBackup added to the comparison.

Some disk space usage statistics with ~= one month of snapshots

HAMMER2 is a filesystem featuring transparent compression and live deduplication.
ext4 is a traditional filesystem with no compression or deduplication features.
Out of the box defaults were used in all cases.