[abconline1]

Hello Everyone,


Just wondering if anyone can suggest me best network design for configuring HA with 8 ESX hosts per cluster (PowerEdge 6850) each having 6 network cards. The client I am supporting is having following configuration -

1. Two Network Cards for vSwitch0 for Service Console and VMotion.
2. Two Network Cards for vSwitch1 for Virtual Machines.
3. Two Network Cards for vSwitch2 again for VMs.

I am seeing lot of issues related to HA & DRS in existing configuration and i am going to suggest following modifications -

1. A backup Service console in each host (is that needs to be in the same network that of primary service console???)

2. A dedicated NIC for service console and one NIC shared betweek backup service console and VMKernel port.


Can anyone give some inputs here?

Thanks in advance!

[jasonboche]

Originally Posted by abconline1

Hello Everyone,


Just wondering if anyone can suggest me best network design for configuring HA with 8 ESX hosts per cluster (PowerEdge 6850) each having 6 network cards. The client I am supporting is having following configuration -

1. Two Network Cards for vSwitch0 for Service Console and VMotion.
2. Two Network Cards for vSwitch1 for Virtual Machines.
3. Two Network Cards for vSwitch2 again for VMs.

I am seeing lot of issues related to HA & DRS in existing configuration and i am going to suggest following modifications -

Originally Posted by abconline1

1. A backup Service console in each host (is that needs to be in the same network that of primary service console???)

Not necessary but new in ESX 3.5.0 is a nagging message in VirtualCenter informing you 24x7x365 that you don't have a redundant COS NIC. The importance here revolves around HA. If the ESX host were to lose COS connectivity, it becomes isolated and hence HA isolation rules will kick in, the default being power off the isolated VMs instantly. You can see how this could be bad should you lose a NIC port on your COS. A secondary COS fixes this single point of disaster.

Originally Posted by abconline1

2. A dedicated NIC for service console and one NIC shared betweek backup service console and VMKernel port.

Your VMKernel port is for VMotion or iSCSI/NAS traffic. Sharing of this NIC assumes you're OK with the sudden burst of VMotion or iSCSI storage traffic on your COS network. It also requires all the VMKernel components to be IPd on the same network. This more or less falls in line with my first comment though.

[abconline1]

Thanks Jason for your reply.

As there are 6 NICs in each host so I think its not a bad idea to have a backup service console on different vSwitch attached to a different NIC.

Also in this case as there will be already a primary service console for the HA hearbeats so I think the vSwitch containing backup service console can be shared with the VMKernel port for VMotion.

Do you know if there is anyway by which we can dedicate a particual service console for HA heartbeats? if that one goes down then heartbeats should start using backup like primary and secondary?