Source Bottleneck

[bertdhont]

Hello,

We have a Veeam Proxy server, connected with a 10Gb ethernet switch.
That proxy server is used for the backup of VMs in vCloud Director.

Those VMs are hosted on ESXi hosts that have a 7Gb uplink (HP blade servers with Virtual Connect) for the management network.

The VMs are stored on a HP Lefthand P4000 cluster (iSCSI). Those storage nodes are on dedicated 10Gb switches.
The ESXi hosts have 2 nics of 3Gb uplink to that iSCSI network.

Transport mode we use is 'Network'.
I know it is not the best mode, but direct SAN isn't possible now (we are working on it to change that).

Now, what I find is strange, is that during backup, the bottleneck is 99% source.
Processing rate is always around 50MB/s, what is very slow...

Anyone having an idea how we can speed up the backup?

Kind regards,

Bert

[dellock6]

Any additional number to better undertand the details? I know almost any component in use, but it's hard to say anything without details. Just as an example, a Lefthand cluster can be designed in a million ways: which kind of disks, how many nodes, what network RAID configuration is in use, etc... Proxy server is a physical machine?