Optimal teaming method for Physical B&R server

[antspants7777]

Hi,

As part of a large VBR implementation, the client is tasked with the physical setup of the server including networking.
The physical Windows 2022 Server has about 100TB of storage and 4 x 10/25Gb NICs (Onboard and PCie)
This will be backing up a 6 node Azure Stack HCI (Hyper-V) environment with a few of VMs over 10TB in size
The switches are enterprise grade Dell switches that are stacked

The client has asked about the optimum teaming configuration.

What is the best practise for the NIC teaming/aggregation?
- LACP
- Switch Independent teaming (SIT)
- Switch Embedded Teaming (SET)

I read that SET is replacing LACP & SIT for virtual environments, but this purely a physical server.
Also SET requires that the make/model be the same, so we would lose NIC redundancy on the VBR server, if we use SET

I understand that most link aggregations are session based, so can only transfer at the 10Gb speed when backing up a single VM.
But, as backups backup from multiple servers at once, we would achieve >10Gb speeds for backups
 
However, would LACP have better throughput for single sessions. E.g. In terms of a single VM restore (RTO)

Thanks in advance

Cheers

Tony

[m.novelli]

Hi Tony, do you have HDD or SSD disks in your physical VBR Server? This is the main repository?

Number of disks / RAID config / controller model?

If you have HDD disk, just one 25 GbE network port streaming from multiple VM should be enough to saturate your backup repository

Remember to install some VM (3 - 4) on Azure Stack side with Veeam Backup Proxy role

Marco

[karsten123]

I would always go for LACP over switch independent if possible.
SET is only possible with Hyper-v rolle installed.
No need for proxies. It is Hyper-v, so its on-host.

[m.novelli]

Ah, nice catch , Hyper-V don’t need proxies!!

Marco