It is considered best practice to separate out various types of network traffic. Usually you would separate out your VM traffic from your management but for this lab we will keep them together. We will separate out vMotion and VSAN traffic though.

ESXi: When I’m designing a vSphere environment with rack mount servers in mind I usually separate management traffic out into a separate standard virtual switch (2 x 1g ports) and all other traffic is sent to 2 x 10g ports through a distributed virtual switch using vlans to separate the traffic further and NIOC to control bandwidth.

All righty then, what’s this going to look like for us?

In Workstation:

So far, so good. For Workstation nothing else needs to be done.

As you can see from the ESXi image above I haven’t specified a physical adapter for vSwitch1, With ESXi if network traffic is on the same VLAN and on the same virtual switch, it won’t go the the physical switch. The virtual switch, an in memory construct, will just pass the traffic along, however if you need to cross VLANs, the traffic will need to be passed to the physical switch for routing across VLANs. In this case, its very handy as we won’t want traffic to pass between the VSAN port group and the vMotion port group. EDIT: What we want to do is set-up the LAN port group with VLAN 4095. This will enable ESXi to pass the vlan traffic about correctly.

Now is ESXi we need to make two changes to the vMotion and VSAN port groups, Enable Promiscuous Mode and Forget Transmits:

William Lam of Virtually Ghetto has a great write up here discussing the reasons for this.

So those are the networking eccentricities, Next we’ll look at getting our first VCSA with a dedicated Platform Services Controller up and running.

Just a note: I had hoped to post this sooner but family and holiday commitments took over.