How to backup a 200tb Windows Server

Can’t help directly but have you though about asking on the Veeam forums? They seem a little more active than here.

What is on the server that you need VSS for? And is this a VMware level backup or Agent level backup you’re trying to achieve?

Hello, I guess this machine could be a File Server? While Veeam Agent for Windows (and any other application leveraging Microsoft VSS) cannot back up volumes larger than 64 TBs, you could protect this server using NAS Backup (a feature of Veeam Backup & Replication). Of course this would be a file-level protection, as opposed to volume-level or machine-level, but it does not have limitations in terms of source volume size.

As a potential additional option, if this is a VM, you could try a crash-consistent backup, disabling both Application-Aware Image Processing and Indexing.

When you set up a backup by default it’ll use VSS.  The server is a stand alone physical Windows 2019 with a large direct attached storage array. 

I’m of the view (without bothering to really research it) that VSS is the process by which changes are tracked.  Without it I think the system would try to backup all the data on each run of the job.  

I have thought about NAS backup but the stupid license calculation means that single backup would cost us about £60,000 in licensing.  400 licenses. 

Well, licensing-wise, yes you need 1 VUL instance every 500 GB of source data. But I suggest you check with your local Veeam Team (Sales + Presales), as there are options with varying levels of discount for large File Shares.

So there aren’t any other options? 

As for now I see three options:

1. Split the volumes into smaller sizes and then back up each volume separately. This would require additional planning and management, but it is a viable option. Sooner or later it becomes an enormous volume with really hard management capabilities.
2. Another option is to use Veeam Agent in “file – level backup mode”, where you can choose the exact directories to be backed up and split the jobs (can take a looong time to complete).
3. And the last one is not to use the Windows Software VSS Provider (that has the actual limitation of 64TB volume for VSS snapshot), but the Hardware one and use the storage snapshots to perform the backup. More details here. Please check if the volume presented to this physical machine resides on the supported storage array.

Are the 200TB on one volume/drive?

Otherwise you can backup with multiple jobs...

It has one drive letter but underneath it is really x2 100TB volumes.  Individually still breaches the 64TB VSS limit. 

I would definitely check out these suggestions for the backup.  Not sure you will find many other options.

It’s bad news though, isn’t it… if we can’t backup more than 64TB with Veeam… unless you break it up or use specific 3rd party technologies.  

If the server is a VM you could back it up with VBR but don't enable Application Aware Processing.  That to me is the only other way around it.  I have seen this work.

This is an interesting idea.  It has the potential to work if what you’ve said about seeing it working is correct. 

What do you think the process is for making this happen? 

I would have to create a VM first of all.  Could I site the VM on the physical server which has the physical disk attached to it?  

By the sounds of it you don’t have a virtual environment so it will take a lot of work to do this if that is the case.  You need to convert the physical box to a virtual one then you could back it up.

I was asking to see if this problem server was already a virtual machine to make it easier.

Naa.  It’s a physical Windows 2019 server.  

I do have places I can create a VM specifically to assist with this backup but you’re not being clear on the details.  Unless you’re saying this only works if the whole server is a VM and the storage is on a SAN or something like that?  

What is the significant difference which means the backup would work if it was a VM type backup? 

What type of files does the server hold and are any of those files super large?

Just wondering if this is critical is it worth looking at rearchitecting it so you can protect it.

e.g. can the data be split across multiple servers or migrated to a new server backed with smaller disks?

Trying to think about restore times etc as well for a server that large.

Yes, my apologies for not being clear.  If this physical box which is Win2019 was a virtual machine on SAN storage you could then take a backup with Veeam without the application aware processing to accomplish your goal.  Sorry for the confusion.

That would be similar to solution 3 from Nikita earlier.  i.e. use the hardware VSS instead of the Microsoft VSS. 

upgrade to V12, use agent with  LVG snapshot

Have you got any more information on the ‘LVG’ snapshot as I can’t find any reference to it.  Have you got a link to the page in the Veeam documentation which describes how to configure this? 

I know @Scott has done backups at this altitude.

We’ve setup a test environment (which can’t be permanent due to licensing) to test the concept of NAS backup in this scenario and we got the following line in the Actions list to show this will likely work. 

“14/03/2023 11:26:01 :: Failed to create a VSS snapshot, failing over to direct backup from the file share Details: Unknown status of async operation
The shadow copy provider had an error. Check the System and Application event logs for more information.
--tr:Failed to create VSS snapshot.
--tr:Failed to perform pre-backup tasks”.  

In other words it attempts VSS as the first method but that fails and it moves to direct mode. 

This is NOT treated as an error so if the backup completes successfully in other respects this line in the backup will not produce a failed or warning status which is good news.  
 

Sorry. I misunderstand you case is Windows backup and think it’s Linux. if Linux, you can do it with LVG snapshot, but no for Windows. Sorry to make you confusion.

As @Jmeixner stated, you can take bakup via multiple backup jobs.  You can use dikpart from windows and split the drives into smaller chunks  ( e.g. 4 * 50TB   or more) and then add each drive letter to different backup job with different backup timings.