Comprehensive and Detailed 150 to 250 words of Explanation From [SR Linux EVPN and Data Center Interconnect/Course Guide/topics]:
Ethernet Auto-Discovery per EVI information is used in EVPN multi-homing to advertise that a PE has reachability to a particular Ethernet Segment for a specific EVPN instance. Remote PEs use this information for aliasing, allowing them to forward known unicast traffic toward a multi-homed Ethernet Segment through eligible PEs rather than relying only on the PE that advertised a specific MAC route. In VXLAN-based EVPN, the update also carries information needed by remote peers to select the proper data-plane encapsulation and VNI for the service. The multi-homing behavior advertised with the Ethernet Segment enables remote peers to understand whether the attachment is operating in all-active or single-active mode. Option D is the false statement in this context because route-target handling is a general BGP EVPN import/export mechanism associated with VPN route policy and extended communities; it is not the specific operational function that defines AD per EVI behavior. The AD per EVI route's purpose is Ethernet Segment reachability for an EVI, not route-target-based service identification by itself. Reference: EVPN route type 1, AD per EVI, aliasing and multi-homing signaling.

Comprehensive and Detailed 150 to 250 words of Explanation From [SR Linux EVPN and Data Center Interconnect/Course Guide/topics]:
In a gateway-less DCI design, there is no dedicated gateway device performing EVPN-to-WAN service interworking. Instead, the data center EVPN overlay is extended across the WAN more directly. Because leaf routers must establish overlay reachability across sites, the IP addresses of the leaf VTEPs need to be reachable through the WAN, commonly by redistributing or otherwise carrying the necessary loopback reachability. The WAN transparently carries the EVPN/VXLAN overlay, and leaf routers can establish VXLAN tunnels across the WAN to remote leaves. Option D is false because it introduces "data center gateway routers" maintaining MP-BGP EVPN peering with the data center route reflector. That is not the gateway-less model; it describes a gateway-based role that does not exist as a separate function in this architecture. In gateway-less DCI, the EVPN control-plane and VXLAN data-plane extension are handled by the fabric endpoints themselves, so the design trades demarcation and interworking control for a more direct overlay extension model. Reference: gateway-less DCI, WAN reachability for leaf VTEPs, transparent EVPN overlay carriage, VXLAN tunnel extension.

Comprehensive and Detailed 150 to 250 words of Explanation From [SR Linux EVPN and Data Center Interconnect/Course Guide/topics]:
An IRB interface is the logical connection point between a Layer 2 MAC-VRF and a Layer 3 IP-VRF. It provides the routed gateway function for hosts in the bridge domain while allowing traffic to move into the routed VRF for inter-subnet forwarding. The correct statement is that when an IRB sub-interface has multiple IP addresses, one can be designated as the primary. This is important because the primary address is used for normal gateway or subnet behavior when more than one address is present on the same routed interface context. Option A is inaccurate because the question is about an IRB sub-interface, not a generic subinterface selection between bridge or routed access modes. Option B is not the defining multi-homing model; EVPN multi-homing is implemented through Ethernet Segment association and MAC-VRF attachment behavior, not by adding multiple IRB subinterfaces for multi-homing. Option D is also wrong because the IRB is the shared logical link between the MAC-VRF and IP-VRF; the design does not require separate IRB subinterfaces on each side as independent constructs. Reference: SR Linux IRB configuration, MAC-VRF to IP-VRF interconnection, primary IP addressing.

Comprehensive and Detailed 150 to 250 words of Explanation From [SR Linux EVPN and Data Center Interconnect/Course Guide/topics]:
Designated Forwarder election determines which PE forwards BUM traffic from the EVPN overlay toward a multi-homed Ethernet Segment for a given service. In this scenario, all leaf routers share the same DF candidate list for mac-vrf103, but the election configuration is not identical. Leaf1 and Leaf2 use the default algorithm, while Leaf3 and Leaf4 use the preference-based algorithm. Under preference-based DF election, the candidate with the highest configured preference is selected over lower-preference candidates, assuming the candidate list is valid and consistent. The question states that Leaf3 has the higher preference value compared with Leaf4. Therefore, Leaf3 becomes the DF for mac-vrf103. This is the correct outcome because the preference-based election explicitly overrides simple default behavior by assigning operator-defined priority to a PE. In production designs, this is useful when the operator wants deterministic forwarding placement, maintenance control, or service-specific primary-path selection rather than relying only on the default modulo-based DF selection process. Reference: EVPN DF election, preference-based algorithm, MAC-VRF service forwarding.