Server backs up extremely slowly via glass and 10g card


I am using a Native Ubackupserver installation on a Cis 1 hardened Ubuntu 22.04 system.
The current installation is UrBackup 2.5.33
I have a shared folder of an NFS share as backup path.
The Urbackup server is installed on an ESXI server as a VM. The network card can handle 10 G. The ESXI server is connected to the Nas via glass.
However, the data transfer is currently 180 kb/s to 2.69 Mbit/s

The Nas has no utilization at all. It is bored.

The Ubuntu Server 22.04 is up to date.
I have noticed that the TOP shows that 1 core always goes to 100% utilization, then the next core goes to 100%. There is no consistent utilization of all cores.

But this is not a condition:

I have not changed much in the configuration, except that there is an nginx reverse proxy in front of it and I use internet clients to encrypt the transmission.

Can anyone help me identify the problem?

Take a look at

I’m no expert on this but, apparently, the client can get into “Paused”. When that happens it seems that UrBackup reports a transfer speed of 50 KBit/s.

Why does it get into Pause? and, when it is paused, why does it show 50 KBit/s implying that there is active traffic between client-server?

I have no answer to those questions but it is worth checking if you client is “paused”.

If your NAS supports docker, install urbackup server on the NAS. This will resolve your performance issues.

You provided too much non-relevant information.

  1. As long as you doing backup to HDD, your 10G network doesn’t matter, your maximum will be near 100Mb/s and up to 200Mb/s synthetically depends on your storage controller performance.
  2. You use VM, but you didn’t show anything about storage setup you use for host and VM, backup drive.
  3. What backup settings you use?

Only thing I can say you use too much layers before target storage, try to reduce them. You use Ubuntu 22.04 which has many obsolet packages, try to run something more fresh like Fedora or Arch with LTS kernel and drivers, because Uroni use this bleeding edge for builds. Experiments with different depending packages can give you some thoughts. You need to tweak systemd try to play with swappiness, choose none as scheduler - can help to avoid some caching issues. Ensure you have something like multipath, jumbo frames, and if possible RDMA for your ESXi - NAS connection. Check Queues, offloadings etc.

This is particular list pissible issues.