Here are 8 key pointers when working with VXLAN BGP EVPN:
1) MP BGP :-
EVPN uses MP-BGP as the routing protocol to distribute reachability information for the VXLAN overlay network, including endpoint MAC addresses, endpoint IP addresses and subnet reachability information.
EVPN is another MP-BGP address family leveraging similar constructs(VRFs, Route Distinguishers (RD) and Route Targets (RT)) as the VPNv4 address family traditionally deployed in MPLS VPN architectures.
EVPN control plane when compared to VPNv4 has the capability of exchanging not only IP but also MAC address information.
2) EVPN Route Types :-
EVPN route update also carry few other updates(L2VNI VNID and VNID on L3VNI tenant VRF, BGP next-hop IP identify originating VTEP, Router MAC of originating VTEP).
3) Endpoint Detection and Tracking :-
A VTEP in MP-BGP EVPN detects attached endpoints via local learning.
MAC addresses are learned in data plane from incoming Ethernet frames whereas the IP address is learned via ARP or Gratuitous ARP (GARP) control plane packets sent by the endpoint.
Once a VTEP detects its local endpoints,it will install a HMM route to track it. It will send an EVPN Type-2 route to advertise the learned MAC and IP address of the endpoint to rest of the VTEPs in same Fabric.
EVPN Type-2 route has an embedded sequence number used for endpoint movement tracking.
4) Arp Suppression & BUM handling :-
Remote Leaf node puts IP-MAC info into remote arp cache and suppresses in coming Arp request for this IP. If IP not present to cache table,VTEP floods the ARP request .
A specific multicast group can be attached to each L2VNI(or shared with all L2 VNIs in a tenant VRF) to replicate BUM traffic.
5) Control Plane :-
BGP neighborship is required between Leaf and Spine. Depends on 1 AS or 2 AS strategy, Leaf and Spine forms IBGP or EBGP neighborship. Spines acting as route-reflector for 1 AS strategy. Address Family EVPN need to use to exchange MAC+IP prefixes.
6) NVE :- Network Virtual Interface (Cisco Terminology).Logical interface where the encapsulation and de-encapsulation occur.
7) Data Plane for L2 VNI :-
A) VLAN to VNI mapping.
B) MAC VRF Config
C) VNI to NVE assignment
8) Data Plane for L3 VNI :-
A) VLAN to VNI mapping
B) MAC-VRF configuration
C) VNI configuration under the NVE interface
L3 VXLAN has following building blocks:
– VRF configuration – RD, RTs, L3 VNI (IP-VRF table)
– Distributed Anycast Gateway for every SVI
– Forwarding SVI for the L3 VNI
– L3 VNI configuration under NVE interface
– advertise prefixes into BGP VPNv4*