I need a plan to create distributed switches in an enviornment that has two different types of servers. One type of server is Cisco UCS with virtual I/O. The other HP servers have physical NICs. The design needed for distributed switches that will be used on hosts utilizing virtual I/O like UCS is different than the design needed for distributed switches that will be used on hosts with physical NICs. However, all hosts need to have access to the same port groups and subnets.
Should I create two sets of distributed switches, one for the UCS servers and one for the HP servers? The same port groups need to be available on all servers. However, the servers with regular 1GB links have special needs that aren't applicable where virtual I/O is in use - such as load based teaming and failover policies that differ from the recommendations for the UCS blades. What is the common way of handling this scenario?
Thanks!