Best local job configuration for off-siting to cloud object storage?

[CaliMSP]

Hi everyone,
I could use some advice on the best way to configure a Veean B&R 11a.


I'm deploying VUL licensed server (non-domain joined stand-alone server+proxies) in an environment that currently runs VBR 9.5 Essentials 3x2 core license (domain joined server + proxies) that does on-site only backups. The goal here is to deploy this new version and configure for off-site backups to Wasabi. New deployment includes Veeam One, which was not a part of the old deployment. At this point I'm considering just stopping jobs on the old server and starting a new, clean chain with this deployment. Old server is running reverse incremental to on-site NAS. I would like to run on-site backups to this NAS and then move backups off-site. After reading the documentation, I'm still confused about the right way to structure this as far as type of the backup job and appropriate settings for off-site.

I understand that to move data to Wasabi S3 compatible storage with immutability, we will need to configure it as a Capacity Tier and set appropriate retention options. What is giving me a pause is what kind of backup job schema is best in this case and how it affects off-loading.

For example this page says that that for reverse incremental https://helpcenter.veeam.com/docs/backu ... ml?ver=110
"all the .VRB files starting from the third restore point will be considered inactive automatically. Thus, you do not need to create an active full (nor synthetic full) backup manually unless you want to offload all the restore points including the most recent .VBK file and the first two .VRB files."
The way i read it is that unless there a separate full backup is made, the VBK (always full, newest backup file) will never be off-loaded to the Capacity tier? If that's the case the cloud storage will only contain incremental, but not full making them useless in case of a major disaster when primary storage is wiped out.

Our full is at 2.5TB right now, with 20-30GB incremental.

So questions are:
1) what on-site backup job type makes the most sense for Capacity tier off-site to S3 compatible storage for this situation? Does this even mater? Can we continue to do reverse incremental on-site?

2) Since the off-site job is triggered by completion of the on-site, i assume after first full it will try to upload this full VBK to to the Capacity tier in the cloud (which given bandwidth limitations will take a few days) . I assume that this active off-site job will prevent the on-site schedule from running until the upload is completed (since the backup file must stay locked and not modified mid-upload).

3) I was previously told that off-load is always incremental forever. So we can set something like 90 days worth of restore points in Wasabi and it will keep one Full (oldest) and 89 incremental (newest). In case of a restore, it will require a pull of all 90 restore points, correct?

[HannesK]

Hell0,

just stopping jobs on the old server and starting a new, clean chain with this deployment.

sounds like a good plan

the VBK (always full, newest backup file) will never be off-loaded to the Capacity tier?

correct. only increments of the reverse incremental chain will be moved.

1) The first question I would ask is whether "move" is the "right" option. I would go for "copy" to protect the entire backup chain in case of a disaster. With "move" and reverse incremental, an outage of the primary repository would make the increments in Wasabi useless. If you only only want to move (no copy), then forward incremental with active / synthetic full is the only thing that makes sense to me (that's personal opinion because I don't want to take the risk of losing all my backups by the outage of one storage system)

2) backup jobs (primary) have preference over offload jobs (secondary). The upload will be continued.

3) Correct, it's incremental forever. But you can configure whatever you like... e.g. 14 days, 4 weeks, 3 months with GFS. I recommend to also look at the sticky FAQ

will keep one Full (oldest) and 89 incremental (newest)

only if you select "copy". The question sounded like "move" to me.

In case of a restore, it will require a pull of all 90 restore points, correct?

restore requires to download the objects that are needed. that's less than the entire chain because the entire chain has all the changes of the last 90 days. And for restore, there is only one restore point that needs to be fetched.

Best regards,
Hannes

[CaliMSP]

Ok that makes sense.

I assume we can do a limited off-load test without immutability flag set and once we see that it's working, then flip the switch on the full backup off-load with immutability?

I'm seeing some conflicting information about adding VMware hosts to a stand-alone Veeam server (one not added to the Windows domain). This document (https://helpcenter.veeam.com/docs/backu ... ml?ver=110) says to add via vCenter, however, I've seen recommendations to add ESXi hosts directly as otherwise processing with hot-add may fail in certain scenarios (backups of the backup server and/or proxies). So is it best to add the vCenter and not backup VBR and proxies, but just do a "configuration backup" from the veeam server? I would assume that configuration backup would not capture the VeeamOne data?

[HannesK]

it depends what you mean with "flip the switch"... in Wasabi or in Veeam? Because bucket versioning must be enabled before doing anything with Veeam. For tests, I would just use an extra bucket.

on the other topic: please create a new topic on that instead of de-railing an object storage forum thread And please provide sources where you have seen that recommendation (because it's wrong). I also recommend looking into the quick start guide

[CaliMSP]

It was here as a way to prevent hot-add failing to network mode. https://www.veeam.com/kb1054