High TCP retransmission rate

[ANorton]

I recently converted over to vDS and I am noticing that I am getting a lot of High TCP retransmission alerts from a lot of the VM's.

Some info about my environment

- VSphere ESXi 4.1 with all updates

- 7 Hosts

- 3 vDS configured as vDS1 is 4 uplink 1gb vnics, vDS2 is 4 uplink 100MB vnics, vDS3 is my ISCSI network, Management networks I left on vSwitch0 for now.

- each host has 4 1gb and 4 100mb links to the vDS it is assigned. The VM's are evenly spreadout over all the hosts.

My questions are as follows:

1. Is this configuration ok or did I mess it up some how?

2. Should I convert all my VM's to VMXNET3 NIC's?

3. Should I be hard setting the NIC's on the actual VM?

Any help would be appreciated.

Thanks

Aaron

[nik-O]

Hi,

A small design question, why did you create several vDS?

[ANorton]

Seperate ISCSI (different subnet), my 1GB vm's and my 100mb vm's. I didn't want to work with vlans to try and seperate everything.

[nik-O]

From a design perspective and to match your objectives, you could create just one vDS and isolate your uplinks as you want.

It's not necessary to create several vDS to do that.

I would be more simple and you could avoid misconfigurations.

[ANorton]

Not sure if I know how to isolate links to different port groups or isolate links to go to certain VM's vs others.