OS10 supports BGP EVPN operation between VLT peers
that you configure as VTEPs. For more information about configurations and best practices
to set up VLT for VXLAN, see Configure VXLAN —
Configure VLT. This information also applies to BGP EVPN for VXLAN.
Dell EMC recommends configuring iBGP peering
for the IPv4 address family between the VTEPs in a VLT pair on a dedicated
L3 VLAN that is used when connectivity to the underlay L3 network
is lost. It is NOT required to enable the EVPN address family on the
iBGP peering session between the VTEPs in a VLT pair because EVPN
peering to the spine switch is performed on Loopback interfaces.
Both VTEPs in a VLT pair advertise identical EVPN
routes, which provides redundancy if one of the VTEP peers fails.
To set up redundant EVPN route advertisement, configure the same EVI,
RD, and RT values for each VNI on both VTEPs in a VLT pair,
including:
In auto-EVI mode, this
identical configuration is automatically ensured if the VNID-to-VNI association is the same on both VTEP peers.
In manual EVI mode, you
must configure the same EVI-to-VNID association on both VTEP peers.
In manual EVI mode, you
must configure the same RD and RT values on both VTEP peers.
In an EVPN configuration, increase the VLT delay-restore
timer to allow for BGP EVPN adjacency to establish and for the
remote MAC and neighbor entries to download by EVPN and install
in the dataplane. The VLT delay-restore determines the amount of
time the VLT LAGs are kept operationally down at bootup to allow the
dataplane to set up and forward traffic, resulting in minimal traffic
loss as the VLT peer node boots up and joins the VLT domain.