Hi,
I have the following scenario. One ESXi with 4 Gbps vmnics. The questions are:
1) If I have a port group configured for "Route Based on the originating Virtual Port ID" in the Load Balancing Policy, and for the same port group I have the option box "Override switch failover order" checked, where I have configured 3 of the vmnics as active adapters and the other remaining vmic as unused adapter, does the ESXi use the load balancing policy that I have configured (in this case "Route based on the originating port ID") between the three vmnics marked as active? Or it does use them in the order that they appear in the active adapters section?
2) Suppossed I have configured the four ports in the physical switch in an etherchannel group in order to use "Route based on IP hash" load balancing policy. In that situation, may I configured for a certain port group to only used two active adapters and the other two ones as unused? In that case, ESXi would load balance using IP hash method but only in the two active adapters? Or it's a misconfigiuration and I shouldn't configure my nic teaming in that way?
3) The official configuration guide says "NOTE IP-based teaming requires that the physical switch be configured with etherchannel. For all other options, etherchannel should be disabled.". How should I configured my virtual networking if I have some port groups that use load balancing policy based on IP hash and another ones use "Route based on the originating port ID". This is the case when I have for example two management ports using the same vSwitch with four vmnics (where they are configured as an Etherchannel in the physical switch). I would have one or more port groups for VMs that use IP hash method for load balancing, and the vmkernel ports por management use only one active adapter with no failback and load balancing as "based on originating port ID" as the best practices says.
Now, the four vmnics are the same for all traffic. The ports in the physical switch must be configured in an etherchannel group because some port groups are gonna use IP hash method, but the other ones don't. The configuration guide says I SHOULD NOT use etherchannel if I'm not gonna use IP hash method, but I WILL use it, but only in one ore more port groups.
Perhaps I cannot share the same vmnics for that situation.
Lastly, this is a philosophical question. What is the difference between the "Route based on the originating port ID" and the "Route based on source MAC Hash" load balancing policy? What is the purpose of the second one? It is supposed that if I had two different MAC address in a VM, that would be because I had two different virtual nics inside the VM, which would be connected to two different port ID in the vSwitch, so I could use the first policy (based on the originating port ID). In other words, which would be the case where I had traffic incoming from the same vSwitch port ID but with different source MAC Address, so I should chose the Source MAC Address load balancing method to discriminate the traffic?
Thanks.
Guido.