The virtual switch concept in virtualization environments is critical to an agile installation. Be sure to get a network admin to check out the configuration, as the configurations could stand to be optimized based on the physical configuration.
As almost every organization has some amount of virtualization, the network standards need a watchful eye by network admins. I mentioned earlier how a co-administered approach is the likely destination for most organization, and now I want to focus on a specific configuration topic with a VMware ESX-based virtual switch. The concept of a virtual switch is a wonderful thing, and VMware’s implementation is solid, because it does not support Spanning Tree Protocol (STP). There is one configuration with the VMware ESX virtual switch where I change the default configuration for load balancing. While most configurations operate correctly with the Route Based on Originating Virtual Port ID option, I use the Route Based on Source MAC Hash option instead. Figure A shows the default option being changed to the MAC hash option.
The reason that I go this route from the default is based on the standard configuration on the corresponding physical switches. With a collective light bulb illuminating, we all now realize that based on the physical switch configuration, we may need to have the same standards apply to the virtual switching. In my physical networking environment, the move to switch communication is based on a MAC hash for load-balancing reasons. Therefore, for consistent configuration in the virtual environment, we have decided to use the MAC hash for load balancing. All functionality is delivered with both configurations; however the MAC hash has a slightly higher overhead on the ESX server. VMware provides two good whitepapers on the network functionality. One is related to the trunking of VLANs on ESX hosts, and the second is related to the virtual switch implementation.