Veeam With vSphere & Nutanix

[DamienL]

Hello,

We have followed the best practice guideline for Veeam Backup with vSphere on one of our Nutanix platform.
However i have a question :

" It’s critically important
to place the ESXi VMkernel port, public CVM interface, public Veeam backup proxy interface,
and Veeam repository (VM or physical) on the same VLAN using 10 GbE. In this example, the
10.10.10.x network is the production management network."

This is our case :
Veeam Backup & Repository is on site B (mgmt network)
Proxy are on site A (dual card, mgmt network & CVM network)
Nutanix full flash platform
Is the veeam repository should have a network interface in the CVM VLAn ?

We are using NFS for backups.
Our repositories are mounted by iSCSI (ReFS 64K), do we need to configure dual card (MPIO) for that ? 2 network interfaces.
Our platform is fully on 10 GBs (prod & mgmt) but i think i have low performance (300 Mb/s W / 500 Mb / R ) (Bottleneck Source most of the time) .
We have adapted also our storage LUN with the right size for Veeam (NLSAS10To 7,2 disk)

Regards,

[Andreas Neufert]

Proxy and Backup Server need access to the Source and vCenter
As well to the Repository.

There are some extra roles for Mount Server (with vPower NFS) that need access to the hosts. If your Repository Server (not the iSCSI backup storage itself) has no access to the ESXi hosts, then you can choose the Proxy server as mount server for the Repository.

More Bandwidth always makes sense. Source Bottleneck can be Source Storage speed, Source network connection or the used backup way.

[DamienL]

Hello Andreas,

As vCenter is in Production network, veeam is also in the production network
Access is working well. I just want to know if the backup server which hosts also the repository role need to have a dual card (1 in prod, 1 in nutanix CVM network) (proxies have dual cards)

[Andreas Neufert]

Now I understand.
Proxy has access to the NFS backend.
VBR Server with Repository do not need access to the NFS network.

[DamienL]

Ok !
But i don't understand why i have a bottleneck with source (FUll flash Nutanix + 10 Gb network access) :/

[Andreas Neufert]

By design there is always something the bottleneck.
To have it at source is ideal, that means your backup environment is fast enough to keep up with your source.
If you are happy with the backup speed then everything is ideal.

[Andreas Neufert]

Can you share the actual Veeam processing numbers not your network numbers?

[DamienL]

Veeam B&R + Repository : 12 CPU / 24 Go
4 Backup repository ( 2 x 2 scale out) : 8 concurrent task per repo
Proxy : Each one : 8 CPU, 16 Go
Concurrent task per proxy : 4

[ronnmartin61]

I think Andreas was looking for backup statistics not compute specifications. It sounds like you're all set up to use DirectNFS but can you confirm that transport method is being used?

[DamienL]

Hello,

OK ! i will give you that.
Yes, we use the recommanded method (direct storage).

[DamienL]

Hello,

Find below an example :

Duration : 57 m
Processing rate : 337MB/s
Processed : 8 TB
Read 250 Go
Transfered : 119 (2,1x)
Bottleneck : Source
Load : Source : 84% > Proxy 34% > Network 26 % > Target 7%

Should i open a case ?

[foggy]

No - as mentioned above, there will always be a bottleneck somewhere in the processing chain, in every setup. That just means that other parts of the chain are doing better than the source. Your speed is pretty good and unless you're unhappy with it - there's no need to open a case.