Virtualized CloudConnnect server with ReFS repository

[akraker]

We just purchased a new higher performance NAS device which will house customer CloudConnect backups. I wanted to take advantage of the space saving and quicker synthetic full backups that ReFS repositories provide.
If nobody sees any issues with this so far my next question is how I should present the NAS storage to the CloudConnect Virtual (hyper-v)? Should I connect the Hyper-V host to the iSCSI target then format as ReFS and create a VHDx file to present to the CC server which will also be formatted as ReFS? Or should I create a virtual network adapter and do iSCSI directly from the VM? Or maybe some other method?

[Mildur]

In your scenario, I would do directly iscsi from the NAS to the VCC VM. Configuring as a hyperV Datastore and using it as a VHDx makes it more complex.

[akraker]

Mildur, That was my initial thought as well. I wasn't certain if running a dedicated NIC in a virtual for iSCSI was supported. With that scenario the networking portion seems a little more complex as opposed to the disk layout being complex.

[Mildur]

The best method for vcc Cloud Connect always will be a physical backup repo (linux or windows). The management vbr and other components can be in a vm.

I have at the moment the same design as you.
I will change that next year, because I see some performance issue if all customers are doing the health check at the same time.
The iscsi+vm backup repo solution doesn‘t fit my scenario anymore. We will go forward with HPE Apollo Linux Hardened Repo Server.

If you can, deploy multiple vms for the veeam component. Calculate the correct RAM and CPU requirements from the deployment Guides.
VCC Gateway (3x best practice)
VCC VBR
VCC Backup Repository

The vcc cookbook is from an old version, but most of the recommendation are true today:

https://vccbook.io/3.Backups/

https://vccbook.io/3.Backups/3.1-backup ... agram.html

[akraker]

With the iSCSI setup I have it should be as close to physical storage performance as you can get without going having a SAN. It is a higher performance NAS with a 10G interface for iSCSI.
Are you saying you are doing iSCSI connected directly from the VM right now? Just curious how you have your Virtual switch setup in your configuration?

[Mildur]

Yes, we have connected our iscsi storage directly in to the vm.
We have two dedicated iscsi vlans with non routable ip ranges. iSCSI Multipath needs that to work correctly.

I don‘t know about the virtual switch configuration. It was done by my colleagues.

My plan for next year is to buy physical hardware backup repos for Veeam Connect Storage (HPE Apolo or something similar).

[akraker]

I have this up and running now. Performance is pretty good with 10g iSCSI between the NAS and the Virtual machine running CloudConnect.
I understand that to benefit from FastClone on CloudConnect jobs I have to do a Active Full backup from each tenant. Will this actually transfer all backup data over the WAN again or does it still utilize change tracking?

[Mildur]

You could also schedule a backup file compact operation. This will activate Fast Clone without an active full.
Active Full will transfer all data again from customer to the Cloud Service Provider.

https://helpcenter.veeam.com/docs/backu ... ml?ver=110

[akraker]

Oh, bummer. I thought I remember doing this once a while back and the active full did scan through the entire backup (not utilizing CBT) but it still only transferred the data that had changed.