borg archive + aws glacier (mirror / reliable backup)

[heapxor]

hello,
i was thinking to mirror borg archive with the most important data to aws glacier (the cheapest storage)
Is it a good idea or does anyone have any view on that? The reason behind is that i have following setup:

laptop/phones etc... to local NAS A;
local NAS A is backed up via borg to NAS B (off site mirror)

So i was thinking to even directly from NAS A backup data to AWS glacier (but e2e encryption has to be done which implies borg)
or just mirror data from NAS B (which contains already borg repo) to AWS glacier.

Any idea / views wellcome.

thanks!

[infectormp]

I think you can find answers in the FAQ
https://borgbackup.readthedocs.io/en/stable/faq.html#can-i-copy-or-synchronize-my-repo-to-another-location
https://borgbackup.readthedocs.io/en/stable/faq.html#when-backing-up-to-remote-servers-do-i-have-to-trust-the-remote-server
and etc.

[heapxor]

thanks, as far as i see its not a good idea to copy repo.. in that case it wont be possible to backup to aws glacier :(

[cyb3rz3us]

I looked at AWS Glacier initially and the only feasible method I found was to use s3fs to attach to an S3 bucket that acts as the Borg repo. It is important to note that s3fs requires closed source code if used on macOS; I mention this because it appears to be a showstopper for some folks.

Also, s3fs can be used on either the client or server side to attach to the Glacier repo; the decision as to which is best would depend on the specific scenario.

One thing to note about Glacier; there is a minimum purchase requirement which is not an issue if you plan on keeping the archives for more than 180 days. If not, then a better solution might be to use a bucket that has Intelligent Tiering which has much more flexible purchase requirements.

In any event, after testing, I decided against using S3 and instead went with an EBS volume attached to an EC2 instance. Yes the cost is much higher than Glacier but it is still comparable to rsync and other Borg-ready services.