An NSX logical switch reproduces switching functionality (unicast, multicast, broadcast) in a virtual environment completely decoupled from underlying hardware. Logical switches are similar to VLANs, in that they provide network connections to which you can attach virtual machines. The VMs can then communicate with each other over VXLAN if the VMs are connected to the same logical switch. Each logical switch has a segment ID, like a VLAN ID. Unlike VLAN IDs, it's possible to have up to 16 million segment IDs.
Before you begin
vSphere distributed switches must be configured.
NSX Manager must be installed.
Controllers must be deployed.
Host clusters must be prepared for NSX.
VXLAN must be configured.
A segment ID pool must be configured.
A transport zone must be created.
About this task
When you are adding logical switches, it is important to have in mind a particular topology that you are building. For example, the following simple topology shows two logical switches connected to a single distributed logical router (DLR). In this diagram, each logical switch is connected to a single VM. The two VMs can be on different hosts or the same host, in different host clusters or in the same host cluster. If a DLR does not separate the VMs, the underlying IP addresses configured on the VMs can be in the same subnet. If a DLR does separate them, the IP addresses on the VMs must be in different subnets (as shown in the example).
- In the vSphere Web Client, navigate to Home > Networking & Security > Logical Switches.
- Click the New Logical Switch () icon.
- Type a name and optional description for the logical switch.
- Select the transport zone in which you want to create the logical switch.
By default, the logical switch inherits the control plane replication mode from the transport zone. You can change it to one of the other available modes. The available modes are unicast, hybrid, and multicast.
The case in which you might want to override the inherited transport zone's control plane replication mode for an individual logical switch is when the logical switch you are creating has significantly different characteristics in terms of the amount of BUM traffic it will to carry. In this case, you might create a transport zone that uses as unicast mode, and use hybrid or multicast mode for the individual logical switch.
- (Optional) Click Enable IP Discovery to enable ARP suppression.
This setting minimizes ARP traffic flooding within individual VXLAN segments---in other words, between VMs connected to the same logical switch. IP discovery is enabled by default.
- (Optional) Click Enable MAC learning if your VMs have multiple MAC addresses or are using virtual NICs that are trunking VLANs.
Enabling MAC learning builds a VLAN/MAC pair learning table on each vNIC. This table is stored as part of the dvfilter data. During vMotion, dvfilter saves and restores the table at the new location. The switch then issues RARPs for all the VLAN/MAC entries in the table.
This example shows the app logical switch with default settings.
- Attach a VM to the logical switch by selecting the switch and clicking the Add Virtual Machine () icon.
- Select the VM and click the right-arrow button.
- Select a vNIC.
Each logical switch that you create receives an ID from the segment ID pool, and a virtual wire is created. A virtual wire is a dvPortgroup that is created on each vSphere distributed switch. The virtual wire descriptor contains the name of the logical switch and the logical switch's segment ID. Assigned segment IDs appear in multiple places, as shown in the following examples.
In Home > Networking & Security > Logical Switches:
In Home > Networking:
Notice that the virtual wires are created on both of the vSphere distributed switches, Compute_VDS and Mgmt_VDS. This is because both of these vSphere distributed switches are members of the transport zone that is associated with the web and app logical switches.
In Home > Hosts and Clusters > VM > Summary:
On the hosts that are running the VMs that are attached to the logical switch, log in and execute the following commands to view local VXLAN configuration and state information.
Displays host-specific VXLAN details.
~ # esxcli network vswitch dvs vmware vxlan list VDS ID VDS Name MTU Segment ID Gateway IP Gateway MAC Network Count Vmknic Count ----------------------------------------------- ----------- ---- ------------- ------------- ----------------- ------------- ------------ 88 eb 0e 50 96 af 1d f1-36 fe c1 ef a1 51 51 49 Compute_VDS 1600 192.168.250.0 192.168.250.1 ff:ff:ff:ff:ff:ff 0 1Note:
If the esxcli network vswitch dvs vmware vxlan command produces the "Unknown command or namespace" error message, run the /etc/init.d/hostd restart command on the host and then try again.
VDS Name displays the vSphere distributed switch to which the host is attached.
The Segment ID is the IP network used by VXLAN.
The Gateway IP is the gateway IP address used by VXLAN.
The Gateway MAC address remains ff:ff:ff:ff:ff:ff.
The Network Count remains 0 unless a DLR is attached to the logical switch.
The Vmknic count should match the number of VMs attached to the logical switch.
Test IP VTEP interface connectivity, and verify the MTU has been increased to support VXLAN encapsulation. Ping the vmknic interface IP address, which can be found on the host's Manage > Networking > Virtual switches page in the vCenter Web Client.
The -d flag sets the don't-fragment (DF) bit on IPv4 packets. The -s flag sets the packet size.
root@esxcomp-02a ~ # vmkping ++netstack=vxlan -d -s 1570 192.168.250.100 PING 192.168.250.100 (192.168.250.100): 1570 data bytes 1578 bytes from 192.168.250.100: icmp_seq=0 ttl=64 time=1.294 ms 1578 bytes from 192.168.250.100: icmp_seq=1 ttl=64 time=0.686 ms 1578 bytes from 192.168.250.100: icmp_seq=2 ttl=64 time=0.758 ms --- 192.168.250.100 ping statistics --- 3 packets transmitted, 3 packets received, 0% packet loss round-trip min/avg/max = 0.686/0.913/1.294 ms ~ #
root@esxcomp-01a ~ # vmkping ++netstack=vxlan -d -s 1570 192.168.250.101 PING 192.168.250.101 (192.168.250.101): 1570 data bytes 1578 bytes from 192.168.250.101: icmp_seq=0 ttl=64 time=0.065 ms 1578 bytes from 192.168.250.101: icmp_seq=1 ttl=64 time=0.118 ms --- 192.168.250.101 ping statistics --- 2 packets transmitted, 2 packets received, 0% packet loss round-trip min/avg/max = 0.065/0.091/0.118 ms
What to do next
Create a DLR and attach it to your logical switches to enable connectivity between VMs that are connected to different logical switches.