This setup looks good but you will want Proxies for the backups as they are the ones that process things for the VBR server.  Depending on your environment if it is VMware you may want to have a Proxy server per host to enable the hotadd feature which speeds up backups.

This site explains proxy servers - https://bp.veeam.com/vbr/2_Design_Structures/D_Veeam_Components/D_backup_proxies/vmware_proxies.html

Hello @Chris.Childerhose,

We are using Hyper-V Environment.
So the location for the Proxies is also okay?

I plan to create a VMs for Proxies.

We have:

1. 16 x Backup Jobs
2. 3 x Backup Copy Jobs over WAN Accelerators
   1. 1 X Backup Copy Job is our biggest VM with 12TB
   2. 1 X Backup Copy Job with 2 VM that equals around 3TB
   3. 1 X Backup Copy Job with 9 small VM’s around 1TB all together.

How many Backup Proxies, and WAN Accelerators would I need?
Those parts are the most confusing ones and I can’t calculate them..

Hi @NemanjaJanicic - yes, your infrastructure looks good as Chris states. Just for clarity, I’d add “Repository” to your Windows VM objects running ReFS/iSCSI in your diagram, as they will be used for the Repositories Veeam component.

To answer your question about Proxies → yes, Proxies are needed. They are a required component for Veeam, as are Repositories, beit the “default” Proxy/Repository (i.e. components on the VBR server itself), or separate servers, as you have. Read more about Proxies here (general purpose) and here (off-host). 

Veeam has a decent diagram themselves in the Guide describing how backup works, which is similar to yours, so you’re looking good there.

And, for your Backup Copy job, yes...you will need a Source and Target WAN Accelerator, so those are good as well. For more on configuring those, see here & here (how they work). Veeam tells how to configure & size them. 

Cheers!

Also, to size your proxies, even though you’re using Hyper-V, use the Veeam Best Practice Guide Chris shared for Proxy sizing. In the Hyper-V section of the Guide, they state to use the VMware section of the Guide which discusses Proxy sizing. Go through it a couple times and you should get the hang of it pretty quick. It’s not too bad.

The info I shared above on Accelerators will help you with sizing those.

Thank you @Chris.Childerhose @coolsport00 

I will soon make the reorganization of everything. Still in the testing phase but it’s already looking really good.


What are the best practices for changing the backup process of everything?
I also need to empty my Sinology NASs to create spaces for iSCSI LUN storage since I’m going to connect them as a repository to our new VMs dedicated only to that.  

What should be some basic testing, how to have some type of Backup when I erase everything and start from scratch? It will be a few days before everything starts going smoothly.

What do you mean "best practice for changing backup process"?

To do a basic test just create a Backup job with a few VMs, enabling the features you want, then see how it performs. 

Yes please clarify what you mean so we are able to give you some tips.  As Shane noted testing it creating jobs, changing options, etc to see how things perform.

I need to reconfigure all our Backup Jobs and backup Copy Jobs from previous configurations to the new ones.

I need to add new WAN Accelerators, Backup Proxies...
I also need to clean our NASs and free up the space so that I can create iSCSI LUN’s and add it to the new Windows Servers that will be connected as backup repositories to the Veeam B&R Server.


How to do it in the best way possible so that our backups are not affected by much. 
How much should I need for all of it

What is the reason for three different backup copy jobs?

Hello @haslund,

The scenario is:
WAN Accelerator 1:
    - 1 X Backup Copy Job is our biggest VM with 12TB
    - 1 X Backup Copy Job with 2 VM that equals around 3TB

WAN Accelerator 2:
    - 1 X Backup Copy Job with 9 small VM’s around 1TB all together.


This is the only difference.
Two Source WAN Accelerators for 3 Backup Copy Jobs.
But nothing is configured “by the books”. I gave 200GB spaces for both Source WAN Accelerators.

Many-to-one is implemented.
2 Source WAN Accelerators on the Primary Site → 1 Target Accelerator on the Secondary Site.

What is your opinion on this?

@NemanjaJanicic: Happy to try and help answer those questions, but let’s first focus on the original question. Why three backup copy jobs? Why not 10? or 2? or 1? How did you decide to make exactly three?

More questions to try and help answer your questions:

1. What is the available bandwidth and latency between the sites?
2. Anything else that needs to use bandwidth during the copy?
3. The size of the VMs are mentioned, but not the daily changes? How many GB/TB of new data do you have today for these VMs?

Thanks for involving in this @haslund.

I will try to answer your questions:

1. Bandwidth between sites should be 1GB. We have a stable connection via IPsec.
2. Veeam Backup Copy Job is set up in the way of Immediate Copying. So, if it's in working hours from 8 am to 4 pm than we have our employees using the IPsec.
3. Daily changes are not that much, to be honest.
   For our biggest VM - Our File Server, Incremental backups never goes over 40-50GB.
    

When are your backup jobs starting? and usually ending today?

@haslund 
Regular Backup Jobs are configured to start at 8PM, 9PM, 10PM.
The smaller ones are relatively quick, the bigger one’s a little bit longer.
Our FileServer incrementals are arroung 45 minutes.

Alright, here are my thoughts on the setup.

Using a virtual machine as the repository server is concerning to me. If someone gains access to your hypervisor host, they can simply mess with things, sure deleting the VM will not delete the backups from the repository, but it sure will complicate things for you. Are you sure you can’t swing a physical server for the repositories even if you still use your Synology via iSCSI? I would strongly recommend this. Try imagining your host fails, you set up a new host but you cannot restore yet because the repository VM must first be deployed and configured etc.

Maybe consider deploying one of the backup repository servers as a Linux hardened repository to gain immutability in your set up?

Using virtual machines for your WAN accelerators is completely fine, but I question if you really need them? By my accounts, your last job should finish around 11 PM or earlier, meaning you have at least 9 hours until business hours. That means you can transfer up to about 4 TB of backups to your off-site location. It sounds like this is a lot more incremental data than you actually need to transfer.

Of course, you can still deploy the WAN accelerators if you really want to - just not convinced there is some real benefit for you. If you do deploy it, make sure to configure it in high bandwidth mode (since your bandwidth is more than 100 Mbit/s) and then there is no need for the cache as it will not be used.

On the backup copy jobs, unless you are changing the compression mode for example then I would probably just consolidate to a single backup copy job to have less management to deal with.

1 ) What type of storage is your primary storage? 

Having a proxy integrate with the storage and perform snapshots takes the load off production, and is very fast.  You would want physical servers to be the proxy in this case.​​​​​​​

​​​​​​​2 ) What made you go with WAN accelerators?

I have many jobs and even single servers over 30TB and don’t use WAN accelerators. With the cost of paying for 2 servers to do that, you could spend the money on a 10Gb connection and be way further ahead.

​​​​​​​

3 ) What is your plan for proxies?

If you go VM’s, you need one on each host for hot add. I run physical and the nice thing is a host can be a proxy, as well as a repository. You could also use it for copy jobs etc. If a proxy is getting busy, just add another server.   If your whole environment crashes, (Hyper-V) you are much better off with physical Veeam infrastructure.

I’ve been in this scenario.  It added a couple hours in my case to get a host online, VM build (add downloading the US ISO if you don’t carry one around in your pocket like I do), then Veeam download, deployment, connect to the repository storage, restore your configuration backup if you have it, rescan the repo and figure out your usernames/passwords if you don’t have them, and THEN you can start restoring.  

I didn’t even notice that. Virtual Machine as a repo is not good.

I totally isolate things, non domain joined, physical hosts.