BGP EVPN control plane

This section discusses the following components of the BGP EVPN control plane.

Supported route types

The following table describes the supported EVPN route types.

Table 1. Supported EVPN route types
Type Description

1

Ethernet Auto Discovery (AD) route

2

MAC/IP advertisement route

3

Inclusive Multicast Route

4

Ethernet Segment Route

5

IPv4/IPv6 Prefix Route

7

IGMP Join Synch Route

8

IGMP Leave Synch Route

Supported BGP functionalities

The following features are supported for BGP EVPN address-family:
  • iBGP and eBGP peering
  • IPv4 and IPv6 peer types
  • Peer groups with EVPN neighbors
  • Ability to negotiate only EVPN address-family for IPv4/IPv6 peers
  • iBGP route reflector server
  • Keeping next-hop unchanged for EBGP peers
  • Ability to retain all EVPN routes without importing them
  • Coexistence of all IPv4 and IPv6 BGP features for neighbors negotiating EVPN address-family
  • EVPN spine, leaf, and border leaf functionalities

Supported data plane encapsulation types

The following data-plane encapsulation types are supported:

MPLS is the default encapsulation type.

The BGP encapsulation extended-community attribute is carried with each route in the BGP control plane to signify the encapsulation to be used to reach the prefix. An EVPN route not carrying any encapsulation extended-community attribute implicitly means that MPLS is encapsulation type.

If an EVPN prefix is received from multiple sources with both VXLAN and MPLS encapsulations, the route source with VXLAN encapsulation is preferred within the data center.

Supported service-interface model

The VLAN-based service-interface model described in RFC 7432 is supported. Each VLAN is mapped to a unique VNI/MPLS label. Within the data center, each VNI maps to a unique EVI; this mapping is referred to as the "single subnet per EVI" option in "draft-ietf-bess-evpn-overlay." No VLAN translation service is supported. Within the data center, VNI-to-VLAN mapping is local to each leaf, and the inner VXLAN frames may not carry an Ethernet tag. The administrator is responsible for keeping this mapping consistent within the data center.