Question posed to our BladeSystem Specialists, "What is the difference between a vNet and a SUS?"

[chuckk281]

The question is confusing, but here is the difference between a vNet and a SUS. When you create a "Network" (vNet) within VC (not a Shared Uplink Set), we can associate one or more uplinks for that vNet, connect servers NICs to that vNet, through the server profiles. A vNet has no concept of VLAN tagging, if the switch that the vNet is connected to is configured for multiple VLANs and/or tagging, those tagged frames will be passed, un-altered, to the server NIC, the server will then need to be able to interpret those tags. If the switch port is configured to NOT tag, then the OS will simply process the packets as normal. When you create a Shared Uplink Set (SUS) we can associate it with one or more uplinks, you then create Networks within the SUS. These Networks are configured with VLAN tags configured. The upstream switch port connected to this SUS will need to be configured for ALL the VLANs configured within the SUS and vice versa. Now, when a server NIC is connected to ONE of these Networks from the SUS, the VLAN tag will be removed before the packet is passed to the server NIC, so the server will not see the tag. In Virtual Connect v1.31, we added the Map VLAN Tags feature, this needs to be turned on, under the Ethernet Settings/Advanced tab, this provides the ability to use a SUS to present multiple "Networks" to a single NIC. In this case, if you select Multiple Networks when assigning a Network to a server NIC, you will then have the ability to configure multiple Networks (VLANS) on that server NIC. At this point ALL packets will be presented to the NIC, tagged, unless the "Native" check box is selected for one of the networks, in which case, packets from this network (VLAN) will be untagged, and any untagged packets leaving the server will be placed on this Network(VLAN). We initially created vNets to present multiple VLANs to a host, IE; a ESX vSwitch. But then created a SUS to present a single untagged network to a Windows NIC or the ESX console NIC. By doing this, the host did not need to understand the tag, as VC took care of it. Now with Mapped VLAN Tags, we can create a SUS, have it contain ALL the VLANs we want to present, the present only ONE network (the one associated with the VLAN we want the server NIC in) to the Windows NIC or the ESX Console NIC, then select MULTIPLE Networks for the NIC connected to the ESX vSwitch and select ALL the networks that we want presented to the ESX host vSwitch. The vSwitch will then break out the VLANs and present them to the guests. The can provide the ability to minimize the number of uplinks required.

[quksilver]

Umm, everything made sense until the Map VLAN tags and Multiple Networks part. Is there a guide with a diagram that describes the difference or the limitations with/without selecting Mapped VLAN and multiple networks?

[chuckk281]

Have you looked at the Virtual Connect Cookbook that is posted on this site in the "Files" section? Let me know if that doesn't help and I will see if our technical guys can help.