Backing up directly to capacity tier B&R v12

[andyboy16]

I see backing up to Capacity Tier is available now in v12. We have a direct connect to AWS and would like to backup directly to S3 bucket. Has anyone utilized this method with B&R v12? If so, any cons to this?

[chris.childerhose]

I have not used AWS but have sent directly to Wasabi without issues, even restores. The only consideration for AWS would be ingress/egress fees.

[RobTurk]

Also consider required restore performance. Do a sanity check on available bandwidth and desired recovery time objective (RTO) if you expect large restores.
A bandwidth calculator may help. Pulling 10TB over a 1Gbit line still takes a day. Therefor a local repository still makes perfect sense for operational backups.

[andyboy16]

I have not used AWS but have sent directly to Wasabi without issues, even restores. The only consideration for AWS would be ingress/egress fees.

Yep, we've considered that. Luckily with the 10GB direct connect fees are reduced.

[andyboy16]

Also consider required restore performance. Do a sanity check on available bandwidth and desired recovery time objective (RTO) if you expect large restores.
A bandwidth calculator may help. Pulling 10TB over a 1Gbit line still takes a day. Therefor a local repository still makes perfect sense for operational backups.

Yep. Definitely considered that but as you know storage is cheap in S3 vs buying new hardware on-prem....which is out of ITs control obviously. Hopefully the 10GB direct connect saves us some time if something happens.

[RobTurk]

When taking everything into account, S3 may or may not be cheaper than on-prem.
A start would be (($/GB/month + gets/puts/scans/GB) * GBs * retention).
Although gets and puts come in per-1000s, it takes many millions of them to transfer large data sets. And retention is recurring cost that can add up significantly.
$1000 / month may sound great, but if your business calls for 10 years retention, that's $120K which buys a very nice on-prem hardenend repository, even considering a hardware cycle after 5 years.

Some select auto-tiering, expecting cost benefit from migrating older data to lower cost tiers. And forget to read what the tiering managent and motion costs per object are.
I've done some calculations and when you take all the fine-print costs into account you might be in for a surprise.

All I'm saying: Just make sure you run all the numbers before committing to a cloud-only repository.

There's also positives to be considered. A full set of backups in S3 may enable fast disaster recovery to EC2, perhaps saving you an entire data center.

[andyboy16]

When taking everything into account, S3 may or may not be cheaper than on-prem.
A start would be (($/GB/month + gets/puts/scans/GB) * GBs * retention).
Although gets and puts come in per-1000s, it takes many millions of them to transfer large data sets. And retention is recurring cost that can add up significantly.
$1000 / month may sound great, but if your business calls for 10 years retention, that's $120K which buys a very nice on-prem hardenend repository, even considering a hardware cycle after 5 years.

Some select auto-tiering, expecting cost benefit from migrating older data to lower cost tiers. And forget to read what the tiering managent and motion costs per object are.
I've done some calculations and when you take all the fine-print costs into account you might be in for a surprise.

All I'm saying: Just make sure you run all the numbers before committing to a cloud-only repository.

There's also positives to be considered. A full set of backups in S3 may enable fast disaster recovery to EC2, perhaps saving you an entire data center.

Ya, I agree, our monthly bill alone just for S3 is well over 3k a month. Can't really justify spending anything for on-prem when we are slowly closing down our datacenters and moving everything in the cloud.
"The cloud is cheaper right?" /s