One ESXi host under performing

[PizzaGuy]

We have installed Veeam Availability Suite 9.5 on a HP DL380 Gen9 running Windows Server 2016. The DL380 is being used as the backup repository and proxy, using Network Mode (NBD), and uses x4 10G SFP cables to connect to our core switch. Our three node ESXi cluster and SAN are also connected the core via 10G links.

I get very consistent 150 MB/s speeds when backing up any VM running on two of the hosts. However, when backing up any VM running on the third host, I get awful spikes and poor processing rates. The bottleneck is Source at 99%, which I see from the FAQ is the first of the four step process, being the initial read from the source. Is there any way to determine if the source is having trouble at a networking, storage or processing level? This particular host is the most robust of the three. vCenter is error free. Is Veeam able to report more detail about the issues with the source to help me troubleshoot?

[Shestakov]

Hello and welcome to the community!

Could you give more details on the backup jobs statistics?
Do you have same bottleneck and transport mode for all 3 hosts?
Generally if you have source as a bottleneck, it`s recommended to change a transport mode to hot-add or direct access. Since you wrote SAN has a connection to the proxy I believe you may leverage the direct SAN mode.

Thanks!

[jmmarton]

If for some reason you can't change the transport mode, take a look at the management interface of the ESXi host exhibiting problems. Perhaps there's an issue there (even something as simple as duplex mismatch).

Joe

[PizzaGuy]

I've found there to be a general error with one of the links to this host. I'll hold off any more additions to this thread until I resolve this issue. Thank you for leading me in the right direction.

[PizzaGuy]

Correcting the issue with the link resolved the original issue. Thanks for the guidance.