Virtual Connect, multiple LOM to same ethernet network

[Bastiaan v H]

Hi,

We have one C7000 with two VC Flex-10. It has two 10Gb LAN access 10Gb uplinks and one 10Gb iSCSI uplink per VC.

I would like to add 4 NIC's to each blade for iSCSI. But i'm not allowed to add more than one Ethernet network instance to a blade. I'm also unable to create a second ethernet network with the same VLAN. So i'm stuck....

So i would like:

Lom1:a: 5Gb / Production & Management

Lom2:a: 5Gb / Production & Management

Lom1:b: 2Gb / vMotion

Lom2:b: 2Gb / vMotion

Lom1:c: 1.5Gb / iSCSI

Lom2:c: 1.5Gb / iSCSI

Lom1:d: 1.5Gb / iSCSI  <= not allowed

Lom2:d: 1.5Gb / iSCSI  <= not allowed

Any idea's how i an achieve this?

[Psychonaut]

Why don't you want to just use two and make the connection 3Gb?

[Bastiaan v H]

The equallogic storage array's have no more then 1Gb interfaces. So our toughts are more NICS are better then only two with high bandwidth. Incorrect ?

[Psychonaut]

With these being CNA's and just one physical NIC underneath the four - having two or four connections going to the same network should get the same results as you have the same physical hardware underneath. From what I've seen there isn't a way around this (maybe with firmware 3.30 and then having a tunneled SUS for those connections?).

My two cents would be to use two connections and set them at 2Gb each (that way you lessen the chance of overloading your uplinks) and bump them to 3Gb if you need to.

[Hongjun Ma]

if your bandwidth bottleneck is on array side, then have 2 3G NICs vs. 4 1.5G from a blade shouldn't make any difference. no any single or aggregate sessions can exceed 1G.

The reason of you can't config ":d" with the same iSCSI vnet name is that VC doesn't allow you config the same vnets off flex NICs within one physical 10G port because you already configured the same name in ":c" flexnic.

but I'll say just make 2 3G NICs for iSCSI.

[Bastiaan v H]

Ok, thank you for clearing this. I will go forward and do some testing with the 2x 3Gb setup.