Offsite Backups to Object Storage

[incognito403]

Hey guys - long time Veeam user first time poster.

I'm testing out using S3 Compatible Object Storage for offsite backups in V10. I'm wondering on the best practice to set this up given my scenario below.

Very simple setup - single backup job with per VM backups. I don't care to "scale out" my backups to S3, I just want a second copy of it in S3 for DR purposes.

I've tested this and it does work but I've had a few issues where certain virtual machines will write the entire VM instead of changes to S3 but this doesn't happen on the local backup only when offloading to SOBR, seems counter intuitive.

How I have it set up is

1 - Performance Tier is using my primary repo as the extent
2 - Place Policy is using Data locality
3 - Capacity Tier is extending to the S3 storage with copy backup jobs to storage as soon as they are created selected and move backups to object storage as they age out is not selected

What is the best setup given my requirement, simply offsite backups of my primary repo as it is to S3 object storage.

[oleg.feoktistov]

Hi Nathan and Welcome to our Community Forums!

I've tested this and it does work but I've had a few issues where certain virtual machines will write the entire VM instead of changes to S3 but this doesn't happen on the local backup only when offloading to SOBR, seems counter intuitive.

I assume, you configured your backup chain without periodic full backups, correct?

If you did, this issue is noted here as a known one. Also, check the related KB article for available hotfix.

As for your setup, I'd suggest using a dedicated gateway server to copy your data to S3 bucket in DR site and look towards Immutability feature in case your object storage provider supports object lock and versioning. Check the most current list of providers here under Compatible + Immutability section.

Also, as a general recommendation, a firewall between Primary and DR sites and ingress/egress traffic restriction for offload/recovery only would be worth implementing.

Thanks,
Oleg

[incognito403]

Thanks Oleg - I saw the hotfix and this does solve the problem.

Another question. It seems when we are running nightly backups on a slow upstream connection that the offload doesn't finish before the next nightly backup, this causes the offload to "reset" and start uploading all over again, which basically means the initial offload never finishes.

Thoughts?

[veremin]

Doesn't look expected, do you mind opening a ticket for this issue, Nathan? Thanks!