Solved

iSCSI ReFS Storage with Synthetic Fulls + Backup Copy Job


Userlevel 7
Badge +1

Hello Veeam Community (probably the best one)

One topic came to my mind today when I was going a little bit deeper into FastClone with ReFS.

The question is this:
On the primary site, Plan is to have VM with ReFS storage connected with iSCSI Initiator to Synology.
With that infrastructure we will enable Synthetic Full Backups which will have FastClone enabled.

Now, because I’m referencing my Backup Copy Jobs to the main Backup Jobs I want to know how this will work in above scenario.

Since FastClone creates Synthetic Full with only references to the previous backups in the chain, what will then I have on my secondary site when that Synthetic Full backup is transferred to the secondary site?

  • Will I have only references to the backups that are located in the Primary site (that is probably something that all of us don't want) since that is what that file is consistent off or it will have references to the already copied Backups on the Secondary site?
  • In case if we lose our Primary Site than our Backups on Secondary Site are corrupted?

 

Thank you all for always engaging and helping.
Best regards,
Nemanja. 😁

icon

Best answer by MicoolPaul 23 August 2023, 10:13

View original

11 comments

Userlevel 7
Badge +20

Hi,

 

Fast Clone is done on a per-repository basis, it leveraging file system functionality so there’s no possibility that Backup on site A references Backup on site B for example. They reference blocks within their own repositories only, but if you’re using ReFS + Fast Clone at both sites, both backup repositories and therefore the independent backup chains within them will both benefit.

 

To explicitly answer your questions:

No you won’t just have references, it’s two independent backup chains, one per site, no block referencing between them.

No if you lose your primary site it won’t impact your secondary site, and vice versa 🙂

Userlevel 7
Badge +17

😁Michael was faster.

I have this scenario running at several customers. Each repository is independent and does not rely on any block references of the other repo.

This works fine for me, space reduction with fast clone is good.

Userlevel 7
Badge +1

Thank you @MicoolPaul @JMeixner for fast reply 🙂

It was a little bit confusing and now its clear 

Userlevel 7
Badge +1

Just to clear things up for me.

Is it better than to also configure Secondary Site to use ReFS with iSCSI instead of SMB?

It’s going over WAN Accelerators at the moment.
 

Userlevel 7
Badge +20

Have you got a server in the second site to connect locally to iSCSI via? I wouldn’t use iSCSI over a high-latency network such as WAN. with SMB where’s the gateway server for SMB located?

Userlevel 7
Badge +1

Hey @MicoolPaul ,

Currently I don’t have the server but I will create an VM with Windows Server and connect iSCSI to it.
After that I would add this server as Backup Repository to our VBR.

Currently on the Secondary Site the Gateway Server for the NAS is Hyper-V Server that is located on the same site.
 

 

Userlevel 7
Badge +1

@MicoolPaul Is this approach good?

To also use ReFS via iSCSI to a VM and use that as Backup Copy repository.

Userlevel 7
Badge +20

Hi, it’s a supported approach that’s fine and I’d always advocate for iSCSI instead of SMB for Veeam.

 

As you already have the hardware I’d just make one suggestion for the future when it comes to inevitably upgrading: I’m not aware of any hardware RAID controllers for Synology NAS Devices currently meaning all RAID is done in software and doesn’t have a battery backed cache. In the future look for either something that supports that or get a lower end server with enough disk enclosure support for the number of disks you need with a battery backed raid controller 🙂 will help protect you from potential data corruption.

 

In the meantime make sure any write cache options are disabled in Synology (unfortunately yes it’s a performance penalty but it means in the event of power loss you’re far less likely to lose data)

Userlevel 7
Badge +1

@MicoolPaul  Thank you for the advices.

I will check how it will affect our data.
I need to check also what will happen with connection to iSCSI when we have an power outage. 
Does it comes back when the power is back, is it automatically connected to the VM etc.

Userlevel 7
Badge +20

You’re very welcome.

 

iSCSI should automatically recover, the other thing you can do is configure Multi-Path (MPIO) on Windows, as your Synology will have multiple network interfaces, you can configure multiple paths to iSCSI this way for redundancy and improved performance depending on the defined policy (policy recommendations typically come from the storage vendors) but Synology don’t recommend one, so I’d just suggest using a load-balancing one for improved performance if you’ve got enough network bandwidth to your VM (e.g. if your server has 10GbE connectivity but your Synology only has 1Gbps)

Synology iSCSI link:

How to Use iSCSI Targets on Windows Computers with Multipath I/O - Synology Knowledge Center

Userlevel 7
Badge +1

@MicoolPaul Thank you so much 🖐

Comment