4A0-D03 Nokia SR Linux EVPN and Data Center Interconnect Questions and Answers

Comprehensive and Detailed 150 to 250 words of Explanation From [SR Linux EVPN and Data Center Interconnect/Course Guide/topics]:

An anycast-gateway IRB allows multiple PEs to present the same default-gateway IP address to hosts in the same subnet. This is the mechanism that enables distributed gateway behavior in EVPN fabrics. The same anycast gateway IP may be configured on equivalent IRB subinterfaces across remote PEs participating in the same IP-VRF, allowing hosts to use the nearest leaf as their default gateway without changing their gateway address. SR Linux also associates gateway MAC information with the IRB, including virtual gateway MAC behavior used inside the MAC-VRF forwarding table. Option C is false because the anycast gateway IP is not treated as a normal unique host route that appears in the IP-VRF route table alongside the subnet prefix. The subnet route is installed for the connected network, but the shared anycast gateway address is a gateway function, not a separately advertised host endpoint that should appear as ordinary routed host reachability. Treating the anycast IP as a regular host route would undermine the distributed gateway model and create ambiguous ownership across PEs. Reference: IRB anycast gateway, MAC-VRF/IP-VRF interconnection, distributed default-gateway operation.

================

Comprehensive and Detailed 150 to 250 words of Explanation From [SR Linux EVPN and Data Center Interconnect/Course Guide/topics]:

Layer 3 EVPN multi-homing with a centralized router uses an Ethernet Segment to associate multiple leaf routers with a common external L3 attachment. The centralized router is part of that attached segment from the forwarding perspective, and the connected leaf routers advertise third-party or customer prefixes into EVPN so that remote leaves can reach those prefixes through the multi-homed attachment. For L3 EVPN, learned customer prefixes are normally advertised using EVPN route type 5, which carries IP prefix reachability. In an all-active design, remote leaf routers may load balance traffic to the customer prefix through multiple attached leaf routers because the ES next-hop allows the remote PE to understand that the prefix is reachable through a multi-homed Ethernet Segment. The false statement is that every router participating in the Ethernet Segment must be configured with all-active mode. Multi-homing mode is a design and configuration property of the EVPN PEs participating in the ES, and designs may use single-active or all-active behavior depending on redundancy and forwarding requirements. Reference: L3 EVPN multi-homing, centralized router attachment, EVPN RT-5 prefix advertisement.

================

Comprehensive and Detailed 150 to 250 words of Explanation From [SR Linux EVPN and Data Center Interconnect/Course Guide/topics]:

In asymmetric EVPN IRB, the ingress PE performs routing from the source subnet into the destination subnet, then forwards the packet across the VXLAN overlay using the destination MAC-VRF/VNI. The egress PE does not perform another IP-VRF lookup for that packet; it performs Layer 2 forwarding in the destination MAC-VRF. In this scenario, Host-2 sends traffic toward Host-1. After the ingress routing decision, the traffic arrives at Leaf-2 in the context of the destination MAC-VRF, mac-vrf-1. Leaf-2 then performs a MAC lookup in mac-vrf-1 and forwards the frame to Host-1. Option D is therefore correct. Option A reverses the forwarding direction and misidentifies the leaf action. Option B incorrectly sends mac-vrf-2 traffic toward the IP-VRF on Leaf-2, even though Leaf-2 is acting as the egress PE for Host-1. Option C is also incorrect because the egress forwarding action is based on the MAC table in the destination MAC-VRF, not an IP-VRF ARP lookup at that stage. Reference: asymmetric L3 EVPN IRB data-plane operation, ingress routing and egress MAC forwarding.

================

Comprehensive and Detailed 150 to 250 words of Explanation From [SR Linux EVPN and Data Center Interconnect/Course Guide/topics]:

A route distinguisher is used in MP-BGP VPN and EVPN address families to make otherwise overlapping tenant routes unique in the BGP control plane. In EVPN, different tenants or EVPN instances may legitimately use the same MAC or IP values. The route distinguisher makes the NLRI globally unique by prepending a unique value to the tenant route. It is typically unique per PE and per EVI, and it is carried in EVPN route advertisements. However, the route distinguisher does not control route import, export, or service membership. That role belongs to the route target, which is a BGP extended community used by receiving PEs to decide which EVPN instance should import the route. Therefore, option D is not part of the correct description of a route distinguisher. Saying that the RD identifies the EVPN instance in the control plane confuses RD uniqueness with route-target membership. The RD makes routes unique; the route target associates those routes with the appropriate MAC-VRF or IP-VRF import policy. Reference: EVPN route distinguisher, overlapping tenant addresses, route target separation.

================

Comprehensive and Detailed 150 to 250 words of Explanation From [SR Linux EVPN and Data Center Interconnect/Course Guide/topics]:

In SR Linux BGP configuration, parameters defined at a more specific hierarchy level can override broader protocol-level settings. Therefore, if a `local-as` value is configured within the BGP group used for EVPN MP-BGP route-reflector sessions, that value overrides the AS number configured directly under the BGP protocol for that group's sessions. Option A is correct. Option B is false because the cluster ID is configured on the route reflector, not on every participating client. The cluster ID identifies the RR cluster and helps prevent route-reflection loops. Option C is false because redundant route reflectors normally operate in parallel rather than as strict primary/backup devices; clients can peer with both for resilience. Option D is misleading because EVPN route reflectors are used to avoid a full mesh of overlay MP-BGP EVPN sessions between leaves, not to replace ordinary underlay eBGP leaf-spine routing sessions. In a clean fabric design, the underlay provides IP reachability, while the EVPN overlay uses MP-BGP sessions, often via route reflectors, to distribute tenant reachability. Reference: SR Linux BGP hierarchy, EVPN route reflector sessions, local-AS override, cluster ID behavior.

================

Comprehensive and Detailed 150 to 250 words of Explanation From [SR Linux EVPN and Data Center Interconnect/Course Guide/topics]:

A gateway-less DCI design extends the data center EVPN/VXLAN model across the WAN without a distinct gateway function separating the data center fabric from the WAN VPN edge. Because the leaf routers or route reflectors must establish the necessary EVPN control-plane and VXLAN data-plane reachability across data centers, the WAN must provide reachability for those fabric endpoints. This design can simplify service continuity and preserve the EVPN overlay model end-to-end, but it does not provide the clean security and QoS demarcation that exists in a decoupled gateway model. A clear demarcation between a border leaf and a separate gateway/WAN PE is specifically a property of decoupled gateway-based DCI, not gateway-less DCI. The gateway-less model also requires reachability between route reflectors or EVPN control-plane endpoints across sites. Therefore, option A is false because it incorrectly assigns the demarcation benefit to the gateway-less design. Reference: gateway-less DCI, stretched EVPN overlay, route-reflector and leaf reachability requirements.

================

Comprehensive and Detailed 150 to 250 words of Explanation From [SR Linux EVPN and Data Center Interconnect/Course Guide/topics]:

This scenario describes all-active Layer 2 EVPN multi-homing with a host connected through a LAG to Leaf1 and Leaf2. The LAG subinterface is associated with the MAC-VRF on both participating leaves, and the Ethernet Segment ES-1 is configured for all-active multi-homing. In all-active operation, both leaf routers can be active attachment points for host-originated traffic, and remote traffic can use EVPN multi-homing mechanisms to reach the segment. Option D is false because the host does not know or use the EVPN Designated Forwarder state when sending BUM traffic. The host forwards over its LAG based on its local LAG hashing and LACP behavior. DF election is an EVPN PE-side mechanism used mainly to control which PE forwards BUM traffic from the EVPN overlay toward the Ethernet Segment, preventing duplicate delivery to the multihomed access network. The host itself does not selectively forward all BUM traffic toward the DF. That distinction is critical: DF controls overlay-to-segment replication, while the host's LAG controls host-to-leaf link selection. Reference: all-active L2 EVPN multi-homing, host LAG behavior, DF election scope, BUM forwarding.

================

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]:

For Layer 2 all-active EVPN multi-homing on SR Linux, the Ethernet Segment configuration is the control-plane anchor that binds the redundant access attachment to EVPN. The relevant LAG subinterfaces must be associated with the Ethernet Segment so that the PE can advertise the segment correctly and apply split-horizon and aliasing behavior. The Ethernet Segment must also be administratively enabled; otherwise, the PE will not participate properly in ES discovery and DF procedures. The multi-homing mode must be set to all-active to permit forwarding through multiple attached leaf routers and support host-side LAG operation. The false statement is option B. The Ethernet Segment Identifier has a defined structure, and the blanket statement that the 2nd through 7th octets “must not be all zeros” is not a valid requirement as stated. What matters operationally is that the ESI uniquely identifies the same multi-homed Ethernet Segment across participating PEs and is consistently configured where required. The ESI must be non-zero as a meaningful segment identifier, but the specific octet restriction in the option is not the SR Linux all-active configuration rule. Reference: SR Linux L2 EVPN all-active multi-homing, Ethernet Segment configuration, LAG association.

================

Comprehensive and Detailed 150 to 250 words of Explanation From [SR Linux EVPN and Data Center Interconnect/Course Guide/topics]:

MAC mobility is the EVPN mechanism used when a host MAC moves from one PE to another. The control plane uses a MAC Mobility extended community and sequence number behavior to determine the most recent valid location for the MAC. When a PE locally learns a MAC that was previously learned through EVPN, it advertises the MAC with an incremented sequence number, allowing remote PEs to prefer the newer location. Therefore, option B is wrong because the sequence number is not decremented. Option A is also wrong because the original PE does not advertise the locally learned MAC with a maximum sequence value as a normal mobility procedure. Option D is inaccurate because PEs do not need direct MAC table synchronization; they rely on EVPN control-plane advertisements and withdrawals. The true statement is option C: the originating PE generates a withdraw message after the same locally learned MAC ages out. This withdrawal removes stale reachability from remote PEs and prevents continued forwarding toward a PE that no longer has the host locally attached. Reference: EVPN MAC mobility, sequence-number handling, MAC route withdrawal after aging.

================

Comprehensive and Detailed 150 to 250 words of Explanation From [SR Linux EVPN and Data Center Interconnect/Course Guide/topics]:

This setup represents single-active Layer 2 EVPN multi-homing. In single-active mode, the Ethernet Segment is configured so that only one PE acts as the active forwarding node for a given service, while the other remains standby. The ports connecting to the host are associated with ES-1 so the EVPN control plane can perform Ethernet Segment discovery, DF election, and standby behavior. If Leaf1 is the active/DF node for the service, all traffic to and from the host flows through Leaf1 until a failure or DF transition occurs. Option D is false because a host LAG is not required for this single-active topology. A LAG is typically required for all-active L2 multi-homing, where the host must treat multiple physical links toward different leaf routers as one logical bundle. In single-active operation, the host can be connected through separate physical links or active/standby access behavior without requiring LACP bundling. The EVPN PEs enforce the active path selection through DF and ES state rather than relying on host-side LAG hashing. Reference: single-active EVPN multi-homing, Ethernet Segment port association, DF-controlled active forwarding.

================

Comprehensive and Detailed 150 to 250 words of Explanation From [SR Linux EVPN and Data Center Interconnect/Course Guide/topics]:

In an SR Linux Layer 2 EVPN MAC-VRF, locally attached MAC addresses are learned through the data plane on local access interfaces, while remote MAC addresses are learned from MP-BGP EVPN control-plane advertisements and installed with a VXLAN next-hop. The exhibit shows one local learned MAC on an Ethernet subinterface and another MAC learned through EVPN with a VXLAN interface and VNI 100. The remote VTEP or next-hop information identifies the remote endpoint, and the VNI maps the received VXLAN traffic to the correct MAC-VRF service. The false statement is C because saying that the MAC address associated with the vxlan-interface “will not age out” is too absolute. A remote EVPN MAC is not aged in the same way as a local data-plane-learned MAC, but it can still be removed when the corresponding EVPN route is withdrawn, invalidated, or no longer present in the control plane. The “N/A” style aging behavior does not mean permanent retention. Reference: SR Linux MAC-VRF verification, local MAC learning, EVPN-learned remote MACs, VXLAN VNI mapping.

================

Comprehensive and Detailed 150 to 250 words of Explanation From [SR Linux EVPN and Data Center Interconnect/Course Guide/topics]:

In interface-less symmetric routing, the EVPN fabric exchanges host forwarding information using EVPN route type 2 MAC/IP advertisements. RT-2 carries the host MAC address and, when present, the associated host IP address, allowing remote PEs to build the forwarding state needed for distributed gateway operation. Unlike asymmetric routing, interface-less symmetric routing does not require every MAC-VRF to be instantiated on every PE. The design scales better because each leaf only needs the locally attached MAC-VRFs plus the shared IP-VRF/routed VXLAN construct for inter-subnet forwarding. Option B describes an asymmetric forwarding pattern more than a symmetric one; in symmetric routing, both ingress and egress PEs perform routed forwarding functions through the IP-VRF. Option C is also incorrect because anycast gateway is fundamental when multiple leaves provide the same default-gateway service for a subnet. Therefore, the true statement is that forwarding information is exchanged using EVPN route type 2 updates. Reference: interface-less symmetric routing, EVPN RT-2 host MAC/IP signaling, distributed IRB operation.

================