Deployment Guide

Each tenant is assigned a VRF and each virtual-network interface is assigned an IP subnet in the tenant VRF. The VTEP acts as the L3
gateway that routes traffic from one tenant subnet to another in the overlay before encapsulating it in the VXLAN header and
transporting it over the IP underlay fabric.
To enable host traffic routing between virtual networks, configure an interface for each virtual network and associate it to a tenant VRF.
Assign a unique IP address in the IP subnet range associated with the virtual network to each virtual-network interface on each VTEP.
To enable efficient traffic forwarding on a VTEP, OS10 supports distributed gateway routing. A distributed gateway means that multiple
VTEPs act as the gateway router for a tenant subnet. The VTEP nearest to a host acts as its gateway router. To support seamless
migration of hosts and virtual machines on different VTEPs, configure a common virtual IP address, an anycast IP address, on all VTEPs
for each virtual network. Use this anycast IP address as the gateway IP address on VMs.
To support multiple tenants when each tenant has its own L2 segments, configure a different IP VRF for each tenant. All tenants share
the same VXLAN underlay IP fabric in the default VRF.
1. Create a non-default VRF instance for overlay routing in Configuration mode. For multi-tenancy, create a VRF instance for each
tenant.
ip vrf tenant-vrf-name
exit
2. Configure the anycast gateway MAC address all VTEPs use in all VXLAN virtual networks in Configuration mode.
When a VM sends an Address Resolution Protocol (ARP) request for the anycast gateway IP address in a VXLAN virtual network, the
nearest VTEP responds with the configured anycast MAC address. Configure the same MAC address on all VTEPs so that the anycast
gateway MAC address remains the same if a VM migrates to a different VTEP. Because the configured MAC address is automatically
used for all VXLAN virtual networks, configure it in global Configuration mode.
ip virtual-router mac-address mac-address
3. Configure a virtual-network interface, assign it to the tenant VRF, and configure an IP address.
The interface IP address must be unique on each VTEP, including VTEPs in VLT pairs. You can configure an IPv6 address on the
virtual-network interface. Different virtual-network interfaces you configure on the same VTEP must have virtual-network IP
addresses in different subnets. If you do not assign the virtual-network interface to a tenant VRF, it is assigned to the default VRF.
interface virtual-network vn-id
ip vrf forwarding tenant-vrf-name
ip address ip-address/mask
no shutdown
exit
4. Configure an anycast gateway IPv4 or IPv6 address for each virtual network in INTERFACE-VIRTUAL-NETWORK mode. This anycast
IP address must be in the same subnet as the IP address of the virtual-network interface in Step 3.
Configure the same IPv4 or IPv6 address as the anycast IP address on all VTEPs in a virtual network. All hosts use the anycast
gateway IP address as the default gateway IP address in the subnet that connects to the virtual-network interface configured in Step
3. Configure the anycast gateway IP address on all downstream VMs. Using the same anycast gateway IP address allows host VMs to
move from one VTEP to another VTEP in a VXLAN. Dell EMC recommends using an anycast gateway in both VLT and non-VLT
VXLAN configurations.
interface virtual-network vn-id
ip virtual-router address ip-address
Configuration notes for virtual-network routing:
• VXLAN overlay routing includes routing tenant traffic on the ingress VTEP and bridging the traffic on the egress VTEP. The ingress
VTEP learns ARP entries and associates all destination IP addresses of tenant VMs with the corresponding VM MAC addresses in the
overlay. On the ingress VTEP, configure a virtual network for each destination IP subnet even if there are no locally attached hosts for
an IP subnet.
• Routing protocols, such as Open Shortest Path First (OSPF) and BGP, are not supported on the virtual-network interface in the
overlay network. However, static routes that point to a virtual-network interface or to a next-hop IP address that belongs to a virtual-
network subnet are supported.
• When you add a static route in the overlay, any next-hop IP address that belongs to a virtual-network subnet must be the only next-
hop for that route and cannot be one of multiple ECMP next-hops. For example, if you enter the following configuration commands
one after the other, where
10.250.0.0/16 is a virtual-network subnet, only the first next-hop is active on the switch.
OS10(config)# ip route 0.0.0.0/0 10.250.0.101
OS10(config)# ip route 0.0.0.0/0 10.250.0.102
VXLAN
11