Hello, I have recently had cause to change one remote backup on s3-compatible object storage from borg to restic, as borg + rclone was using sufficient bandwidth to generate bandwidth charges which $work (a small company) is keen to avoid.
Restic's native s3 compatibility (and possibly reduced metadata churning) seems to have reduced overall bandwidth usage by over 60%. I'm inclined to think the provider (katapult.io) may have some issues/inefficiencies as borg backups took upto 6x as long in comparison to the same process at digital ocean, which they have been looking into for months. The steep drop in bw usage resulting from the switch (according to vnstat) is surprising as backups to that provider represent less than 50% of all activity from the vps concerned. I have only once had an (unrelated) borg repo become corrupted, and from reading around the subject, newer versions of Restic seem to be much more robust than earlier ones. I would be interested to hear from anyone with any insight into restic issues in particular (restore difficulties?) I gather Mount is better than native restore/extract for large files/repos. Any other issues to be aware of on non-Amazon s3? Many thanks, Gareth
