Offsite backup device, Exagrid? Linux Server? Windows Server?

[HendersonD]

Our production data center has 40 VMs being housed on a Nimble Array. We are using Veeam to backup these VMs to a DR site across campus with the target being another Nimble array. We have an older Exagrid box sitting at our ISP 20 miles away. We run a second backup job with the target being the Exagrid. A backup copy job cannot use the Exagrid as a target.

Given the age of the Exagrid we are thinking about replacing it with

1. A new Exagrid
2. A Windows server with lots of internal drives
3. A linux box with lots of internal drives

Any suggestions?

[nitramd]

Do you need significant deduplication?

Windows or Linux will serve you well as a data repository. You could use ReFS with Windows but I would suggest that you check out the ReFS posts here if you haven't already.

[HendersonD]

In my DR site on the Nimble Array I have a lun that is presented to my physical Veeam proxy server and I am using ReFS for it. I have been backing up using ReFS for several months now and am happy with the setup. If I do purchase a new server with lots of internal storage, and install Server 2016 or 2019, isn't there deduplication I can turn on?

[HendersonD]

In other words can I

1. Purchase a server with lots of internal storage
2. Install Windows Server 2016 or 2019
3. Make a share on this server using ReFS
4. Make the share the target of a Veeam Backup Copy job
5. Turn on deduplication under Windows

The only thing that worries me about this approach is looking at the 3-2-1 rule for backup

• Keep 3 copies of your data - I have this with one in production, one on my Nimble Array in my DR site, and one offsite at my ISP
• Store 2 backup copies on different devices or storage media. My original data in many cases is on a Window server. I am using a Veeam physical proxy server with a ReFS volume for backup and if I go the route described above I have another Windows server with a ReFS volume offsite. Using all Windows in my backup chain makes me nervous
• Keep 1 backup copy offsite - this I have

[nitramd]

Veeam does a small amount of duplication but nothing like ReFS or ExaGrid, AFAIK.

I concur with your strategy of diversifying your backup chain. So, this leaves ExaGrid or a Linux repo with no further dedupe (I think).

If I remember correctly, ExaGrid allows you to "continually" add newer hardware to your cluster, i.e. continue to run old hardware until it breaks and can't be fixed as no parts are available; the landing pad sounds nice to have.

If you go with Linux, try the XFS filesystem - it's designed to support a maximum file system size of 8 Exabytes but the limitation here is typically the OS. As for hardware, I would suggest a Cisco S3260 - they're quite capacious.

Of course, our budget pretty much determines what we get

Hope this helps.

[foggy]

A backup copy job cannot use the Exagrid as a target.

I'm a bit confused by this, where does this statement come from?

[*]Store 2 backup copies on different devices or storage media. My original data in many cases is on a Window server. I am using a Veeam physical proxy server with a ReFS volume for backup and if I go the route described above I have another Windows server with a ReFS volume offsite. Using all Windows in my backup chain makes me nervous

Having both repositories running Windows is ok, this point is still fulfilled by the fact that you have two different systems, one of which should cover you in case the other is down. This protects you from losing all the backups if they are stored on a single device and this device goes off.

There are a couple of existing topics regarding ReFS + Windows deduplication usage, here's one of them FYI.