> On 5 Mar 2026, at 13:54, Bigsy Bohr <[email protected]> wrote:
>
> On 2025-12-21, Gareth Evans <[email protected]> wrote:
>>
>>> bandwidth usage (and thus cost, for commercial S3 backends) is likely
>>> due to the increased amount of client-side caching of repo data that it
>>> does compared to borgbackup.
>>
>> Thanks Andy. Restic has resulted in much reduced bandwidth in
>> comparison to borg + rclone (where I gather rclone is the culprit) but
>> is, as you say, reported to be less efficient on that front than plain
>> Borg.
>
> I think he said the exact opposite.
Yes, I had my sources confused.
> I just started using restic on an S3 backend. Easy to set up.
> I like how you can mount snapshots locally and browse at your leisure.
Borg offers this but backups (inter alia) fail if performed while the repo is
mounted.
Rclone + borg seems to absolutely hammer bandwidth.
Restic seems the better solution at least for S3 purposes, and seems to be much
faster than borg+rclone. Borg2 appears to have S3 compatibility in the
offing...
Thanks,
Gareth
>
> Of course, if you can't restore reliably, all bets are off.
>
>> Your experience is reassuring.
>
>
