I have 3 VMWare ESX 4.1.0 874690 hosts each with 8 physical interfaces. 4 Broadcom PowerEdge R710 BCM5709 GigE and 4 Intel 8271EB GigE.
The current configuration is not utilizing all interfaces and we have one Cisco 3750g 24 port switch. This switch is in the same rack as the ESX hosts and or EMC Celerra, so all is connected to it. 2 ports on here are trunked to the core.
We are going to throw in another switch for redundancy. Its a Cisco 3560x. Newer switch, dual removable power supplies, just doesn't have the stacking port. It is in the VM and EMC rack now and it also has 2 ports trunked to the core, but that's it for now.
I'm trying to determine the best way to connect the ESX servers across these two switches to make it redundant. I'm counting my interfaces and I just keep running out of nics to do what I want.
First question,
I have a vSwitch0 which is vmnic0 (broadcom) and it is belonging to "Service Console". It has an ip in vlan1. Does this interface have to be on its own virtual switch? Could the service console be on the same switch as "LAN" traffic considering we use vlan1 for servers "lan" connection.
vswitch1 has vmnic6 (intel) and vmnic2 (broadcom) trunked in VLAN10 "Storage Network" On here is a vswif1 and vmk0 interface both on the subnet that is dedicated to NFS or iSCSI.
vSwitch2 is the DMZ network (VLAN20) and is attached to vmnic5. vmk1 with an ip of 1.1.1.x is also on here since the DMZ is light on traffic, but it has that IP that is completely different from the subnet of the regular DMZ.
vSwitch3 is the LAN network and has vmnic7.
I was thinking of doing this
switch 1 primary connect these
vmnic0 (broadcom) and vmnic4 (intel) trunked as LAN network
vmnic1 (broadcom) and vmnic5 (intel) trunked as SAN network
vmnic2 (broadcom) on DMZ network
switch 2 failover connect these
vmnic3 (broadcom) on LAN network
vmnic6 (intel) on SAN network
vmnic7 (intel) on DMZ network
But now I've run out of interfaces for the service console, hence my main question above.
Next question, would this proposed way be ok you think? Also if something happened to switch1, would vm's and ESX hosts automatically utilize switch2? I realize on switch2 they are not trunked, but this is like warm dr... running on a spare tire per say. Your not down, things are up.. .maybe a little slower but that's better than nothing while you fix the issue with switch 1.